From dbdf56e99c598c81c55ec91ee6668b440bc6d03f Mon Sep 17 00:00:00 2001
From: "flashduty[bot]" <bot@flashduty.com>
Date: Fri, 9 Oct 2026 08:27:38 +0000
Subject: [PATCH] =?UTF-8?q?docs:=20doc-review=202026-10-09=20(diff)=20?=
 =?UTF-8?q?=E2=80=94=20dashboard=20time=20variables=20&=20display=20time?=
 =?UTF-8?q?=20zone,=20RUM=20Explorer=20analytics,=20MCP/Go-SDK=20OAuth,=20?=
 =?UTF-8?q?CLI=20--timezone?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 docs.json                             |   6 +-
 en/ai-sre/automations.mdx             |   7 +-
 en/ai-sre/im.mdx                      |   6 +-
 en/ai-sre/mcp.mdx                     |   6 +-
 en/ai-sre/memory.mdx                  |   2 +-
 en/changelog/changelog.mdx            |  23 +++++
 en/developer/cli.mdx                  |  26 ++++-
 en/developer/go-sdk.mdx               |  14 ++-
 en/developer/mcp-server.mdx           |  44 +++++++-
 en/monitors/dashboards/dashboards.mdx |  38 ++++++-
 en/monitors/explore/explore.mdx       |  27 ++++-
 en/openapi/go-sdk.mdx                 |  12 ++-
 en/rum/explorer/analytics.mdx         | 139 ++++++++++++++++++++++++++
 en/rum/explorer/overview.mdx          |   6 ++
 en/rum/quickstart/app-management.mdx  |   8 +-
 zh/ai-sre/automations.mdx             |   7 +-
 zh/ai-sre/im.mdx                      |   6 +-
 zh/ai-sre/mcp.mdx                     |   6 +-
 zh/ai-sre/memory.mdx                  |   2 +-
 zh/changelog/changelog.mdx            |  23 +++++
 zh/developer/cli.mdx                  |  26 ++++-
 zh/developer/go-sdk.mdx               |  14 ++-
 zh/developer/mcp-server.mdx           |  44 +++++++-
 zh/monitors/dashboards/dashboards.mdx |  38 ++++++-
 zh/monitors/explore/explore.mdx       |  27 ++++-
 zh/openapi/go-sdk.mdx                 |  12 ++-
 zh/rum/explorer/analytics.mdx         | 139 ++++++++++++++++++++++++++
 zh/rum/explorer/overview.mdx          |   6 ++
 zh/rum/quickstart/app-management.mdx  |   8 +-
 29 files changed, 676 insertions(+), 46 deletions(-)
 create mode 100644 en/rum/explorer/analytics.mdx
 create mode 100644 zh/rum/explorer/analytics.mdx

diff --git a/docs.json b/docs.json
index 373594a42..a7da9cf86 100644
--- a/docs.json
+++ b/docs.json
@@ -3728,7 +3728,8 @@
                 "icon": "magnifying-glass",
                 "pages": [
                   "zh/rum/explorer/overview",
-                  "zh/rum/explorer/data-query"
+                  "zh/rum/explorer/data-query",
+                  "zh/rum/explorer/analytics"
                 ]
               },
               {
@@ -5410,7 +5411,8 @@
                 "icon": "magnifying-glass",
                 "pages": [
                   "en/rum/explorer/overview",
-                  "en/rum/explorer/data-query"
+                  "en/rum/explorer/data-query",
+                  "en/rum/explorer/analytics"
                 ]
               },
               {
diff --git a/en/ai-sre/automations.mdx b/en/ai-sre/automations.mdx
index 644ffa5bb..473396466 100644
--- a/en/ai-sre/automations.mdx
+++ b/en/ai-sre/automations.mdx
@@ -58,6 +58,7 @@ The configuration form has the following fields:
 | Scope | Yes | Use the **scope selector** to choose **Personal** (`team_id=0`) or a **Team** (`team_id>0`). Scope determines both the rule's ownership and edit permissions, and constrains the self-hosted Runners selectable in **Environment** — only account-global Runners and Runners belonging to the same team as the scope are selectable. |
 | Environment | No | Use the **environment picker** to choose the runtime environment: **Auto** (the backend picks the best available environment, the default), **Cloud Sandbox**, or a **self-hosted (BYOC) Runner**. After picking a team scope, a team Runner that does not belong to that scope is cleared automatically. |
 | Task prompt | Yes | Describe the task for AI SRE to perform, written in a rich-text editor. This prompt is exactly what is sent to the agent on each run. Placeholder: `Describe the task for Flashduty AI SRE to perform.` |
+| Time zone | No | A field on the **Schedule** trigger card (see "Schedule (cron)" below): the **rule time zone** the cron expression is read in. Pick an IANA zone name (for example `Asia/Shanghai`) from a searchable dropdown; the options are ordered by their UTC offset. On create it defaults to your effective time zone (member time zone first, falling back to the account time zone); when no zone is available the field is left empty and the create request omits it. |
 
 <Note>
 For **Environment**, "Auto" has the backend pick the best available environment at each run; "Cloud Sandbox" is a platform-managed ephemeral sandbox; a self-hosted Runner runs on your own machine. See <a href="/en/ai-sre/environments">Environments</a> for the differences and how to connect them.
@@ -96,13 +97,17 @@ To avoid hand-writing the expression, the UI offers four modes:
 | Custom | Type a 4-field or 5-field cron expression directly |
 
 <Warning>
-**Time zone**: in **Daily / Weekly** modes, the time you pick is interpreted in your **local time zone**, converted to UTC on save; the UI labels your local time zone next to the cadence summary. In **Custom** mode the expression is interpreted in **UTC**, labeled `UTC` in the UI.
+**Time zone**: a **Schedule** run is read against the cron's **rule time zone** — pick it from the **Time zone** dropdown on the trigger card. All four modes (**Hourly / Daily / Weekly / Custom**) share that one field; there is no split between "this mode uses your local zone, that one uses UTC". Saving stores the IANA zone you picked as-is, no longer converting the hour and minute to UTC, and the cadence summary no longer carries a time zone label beside it.
 </Warning>
 
 <Note>
 The actual execution time may differ from the set time by a **few minutes**. This is intentional, to spread out system load. Do not treat a rule as a second-accurate timer.
 </Note>
 
+When you edit a rule that is **enabled** and already has its next run scheduled, changing the cron expression or the **time zone** adds a note to the trigger card: **the occurrence that is already scheduled and has not started yet (`<time>`, `<old zone>`) will not run. The next one is recalculated on the `<new zone>` clock, and its real start can be up to 5 minutes later than that minute.**
+
+The note appears only when the schedule really changed: switching the time zone between an empty value and `UTC` does not count (they are the same clock face), and it does not appear when the **Schedule** trigger is off or the rule has no pending run.
+
 ### Call via API (HTTP POST)
 
 Lets you trigger this automation on demand from an external system, independent of a time cadence.
diff --git a/en/ai-sre/im.mdx b/en/ai-sre/im.mdx
index f5a6684b1..c13319a41 100644
--- a/en/ai-sre/im.mdx
+++ b/en/ai-sre/im.mdx
@@ -83,13 +83,13 @@ The **reply mode** is configurable (off / first / all), controlling whether AI S
 
 ### Processing Receipt
 
-Once the bot has your message, it first puts an emoji reaction on **the message you sent**, as an immediate "received, working on it" signal. The emoji is picked **at random** from that platform's own pool rather than being fixed:
+Once the bot has your message, it first puts an emoji reaction on **the message you sent**, as an immediate "received, working on it" signal. The emoji is picked **at random** from that platform's own pool rather than being fixed — the pool keeps only glyphs that read as "received, working on it", and congratulatory, thumbs-up, heart, or smiley reactions are left out because on a message that just arrived they read as a verdict or as sarcasm:
 
 | Platform | Emoji pool (one picked at random) |
 | --- | --- |
 | Slack | eyes, thinking_face, hourglass_flowing_sand, mag, zap |
-| Lark / Feishu | Typing, Thinking, Get, CheckMark, OnIt, Hundred, and more |
-| DingTalk | Platform built-in emotions: 思考, 收到, Get, 赞, 比心, OK, and more |
+| Lark / Feishu | Typing, THINKING, Get, OnIt |
+| DingTalk | Platform built-in emotions: 思考, 收到, Get |
 
 On Slack this sits alongside a native "is working on it" status, which the bot also shows.
 
diff --git a/en/ai-sre/mcp.mdx b/en/ai-sre/mcp.mdx
index 0cea0ea0a..cee49789c 100644
--- a/en/ai-sre/mcp.mdx
+++ b/en/ai-sre/mcp.mdx
@@ -55,7 +55,7 @@ The **Recommended** block above the catalog ("Recommended from your alert source
     The dialog first shows the template's details: the vendor ("Official Datadog · MCP" or "Community · MCP"), a link to its official docs, its description, and a set of "**Ask things like this once connected**" example questions. Then come the connection parameters the template declares — URL / endpoint placeholders and environment variables. **Shared** fields (shared account-wide under service authorization) are labelled "Shared by the whole account."; in **Member authorization (key)** mode there is also a "**Your API token**" field (a password input) labelled "Only for your own use — other members fill in their own the first time they use it.", with a "How to get one" docs link beside it.
   </Step>
   <Step title="Pick an execution environment and submit">
-    Choose the **execution environment** this connection should be made from — "Connects in the session's own environment by default."; for a template that needs a self-hosted Runner, the hint reads "If 〈name〉 is only reachable on your internal network, pick a self-hosted Runner on that network." The primary button is **Connect**; for a **Member authorization (OAuth)** template it reads "**Authorize 〈name〉**" and creates the connector first, then opens the browser authorization window (OAuth discovery, Dynamic Client Registration, and token exchange all run from the chosen environment).
+    Choose the **execution environment** this connection should be made from — "Connects in the session's own environment by default."; for a template that needs a self-hosted Runner, the hint reads "If 〈name〉 is only reachable on your internal network, pick a self-hosted Runner on that network." The primary button is **Connect**; for a **Member authorization (OAuth)** template it reads "**Authorize 〈name〉**" and creates the connector first, then opens the browser authorization window (OAuth discovery, Dynamic Client Registration, and token exchange all run from the chosen environment; when the authorization server does not support DCR, fc-safari falls back to a Client ID Metadata Document — see "Authentication" below).
   </Step>
   <Step title="Confirm the result">
     On submit, the system creates the connector and runs one connection test against it. On success the dialog shows "**〈name〉 connected**" plus the example questions: click **Done** to close it, or click an example question to open a new chat with that question **prefilled** (it is not sent automatically). On failure it shows "**The connection test failed**" with the raw error: click **Retry** to re-run the test; for a template that needs a Runner, a non-cloud environment also offers a "Let AI SRE install it on the Runner" entry, which opens a new chat that carries no token and asks the agent to install the missing command-line tools (`npx`'s Node.js, `uvx`'s uv) on the self-hosted Runner.
@@ -158,7 +158,7 @@ MCP servers support three **authorization modes** that determine how credentials
     Suitable for third-party SaaS where each engineer uses their own API key (e.g., a personal GitHub token).
   </Tab>
   <Tab title="Member authorization (OAuth)">
-    Each member authorizes individually via **OAuth 2.1**. An authorization window pops up automatically the first time a tool on this server is invoked. Server discovery and Dynamic Client Registration (DCR) are performed as needed; tokens are stored per member and refreshed automatically near expiry. Suitable for services that support OAuth.
+    Each member authorizes individually via **OAuth 2.1**. An authorization window pops up automatically the first time a tool on this server is invoked. Server discovery and client registration are performed as needed: when the authorization server advertises a registration endpoint, fc-safari uses **Dynamic Client Registration (DCR)**; when there is no registration endpoint but the server declares **Client ID Metadata Document** support, fc-safari registers no client and instead uses the `/safari/oauth/client-metadata.json` document it publishes as its `client_id`, leaving the authorization server to fetch it by that URL — nothing extra to configure on your side. Tokens are stored per member and refreshed automatically near expiry. Suitable for services that support OAuth.
   </Tab>
 </Tabs>
 
@@ -180,7 +180,7 @@ MCP servers using service authorization have no per-member authorization step an
 
 **Actions in the Credential dialog**: the dialog offers actions based on its authorization mode and your current credential state:
 
-- **Member authorization (OAuth)**: first choose an **execution environment**, then click **Authorize** (when disconnected) / **Reauthorize** (when connected or expired). Choose a Cloud Sandbox or an online BYOC Runner; **Auto** is not available. OAuth discovery, Dynamic Client Registration, token exchange, and later refreshes all run from that environment. On each open, the dialog prefers the last successful environment if it is still online, then an online runner bound to the connector, and finally the Cloud Sandbox. For an OAuth service reachable only on a private network, choose a BYOC Runner that can reach it. A browser popup then opens the authorization window (initiated via `/safari/credentials/oauth/initiate`, which returns an authorize URL opened in the popup).
+- **Member authorization (OAuth)**: first choose an **execution environment**, then click **Authorize** (when disconnected) / **Reauthorize** (when connected or expired). Choose a Cloud Sandbox or an online BYOC Runner; **Auto** is not available. OAuth discovery, Dynamic Client Registration, token exchange, and later refreshes all run from that environment (when the authorization server does not support DCR, fc-safari falls back to a Client ID Metadata Document — see "Authentication" above). On each open, the dialog prefers the last successful environment if it is still online, then an online runner bound to the connector, and finally the Cloud Sandbox. For an OAuth service reachable only on a private network, choose a BYOC Runner that can reach it. A browser popup then opens the authorization window (initiated via `/safari/credentials/oauth/initiate`, which returns an authorize URL opened in the popup).
 - **Member authorization (key)**: click **Enter Key** (when disconnected) / **Update Key** (when connected) to open the secret-entry modal, which submits to `/safari/credentials/secret`.
 - **Revoke**: when a credential exists, click **Revoke** (via `/safari/credentials/revoke`) to delete your saved credential; the status returns to Disconnected.
 
diff --git a/en/ai-sre/memory.mdx b/en/ai-sre/memory.mdx
index 8a559c2ae..36dcc3d40 100644
--- a/en/ai-sre/memory.mdx
+++ b/en/ai-sre/memory.mdx
@@ -74,7 +74,7 @@ Go to **AI SRE → Customize → Context → Memory**.
 
 - **Pick a scope**: switch between **Personal** and **Team** at the top; for Team, pick one of your teams from the dropdown. There is no Shared (account-level) memory. The scope and team are kept in the page URL, so a refresh or a shared link opens the same set of memories.
 - **Find a memory**: the list on the left is grouped by Preference / Procedure / Fact / Lesson, newest update first within each group; the search box filters by name, description, and file name.
-- **Create or edit**: click **New memory**, or click **Edit** on the right, fill in the name, type, a one-line description (when it applies), and the content, then save.
+- **Edit**: click **Edit** on the right, change the name, the one-line description (when it applies), and the content, then save. This page has no create entry — memory is produced only in a conversation (see [How Memory Is Created](#how-memory-is-created)); the **type** cannot be changed here either, because it is fixed when the memory is written. The page only groups by type and writes the type back unchanged.
 - **Delete**: after deletion, later conversations no longer use it. Telling the assistant to forget something in the conversation has the same effect.
 
 If someone else or the AI changed the memory while you were editing, saving warns you; saving again overwrites it with your version, and canceling shows the latest content.
diff --git a/en/changelog/changelog.mdx b/en/changelog/changelog.mdx
index 8b3668059..0000b1ec0 100644
--- a/en/changelog/changelog.mdx
+++ b/en/changelog/changelog.mdx
@@ -4,6 +4,29 @@ description: "This page documents important updates and feature releases for Fla
 keywords: ["Changelog", "Product Release", "Feature Updates", "Flashduty", "Version History"]
 ---
 
+<Update label="2026-10-09" description="📊 Dashboard built-in time variables and display time zone, RUM Explorer analytics, OAuth for MCP Server and the Go SDK">
+
+### Dashboards: built-in time variables and display time zone
+
+Panel expressions can now use the built-in time variables `$__interval`, `$__rate_interval`, `$__range` and `$__auto`, rendered into seconds from the panel's own execution window and step — writing `$__range` removes the need to hand-maintain a variable just to know how long the window is. Availability depends on the datasource and query mode: Prometheus range queries support the first three, Loki range queries support `$__auto`, `$__interval` and `$__range`, Loki instant queries support `$__auto` and `$__range`, and any other combination is rejected on save. A leading `__` is reserved for built-in variables, so custom variables can no longer use that prefix.
+
+The time picker footer gains a **Display time zone** row: switch between **Default** (follow your effective time zone), **Browser time**, **Coordinated Universal Time**, and a city list grouped by continent. It applies to dashboards only — it drives how the time picker displays and parses time and how the time-series x-axis and tooltips read — and the choice is remembered in this browser. See [Dashboards](/en/monitors/dashboards/dashboards).
+
+### RUM Explorer: analytics view
+
+The Explorer gains **List** and **Analytics** tabs. The **Analytics** view aggregates the events of the current event type by aggregation (count, distinct count, average, sum, minimum, maximum and P50 / P75 / P90 / P95 / P99) and group field, and offers five chart types: time series, top list, pie, table and distribution. You can drill from a chart element back into the list, drag-select a time range to zoom, download a PNG or export CSV; the whole analytics state lives in the URL, so copying the link reproduces the same analysis. See [RUM Explorer analytics](/en/rum/explorer/analytics).
+
+### OAuth for MCP Server and the Go SDK
+
+The remote MCP service `https://mcp.flashcat.cloud/mcp` now accepts both APP keys and **MCP OAuth access tokens**: with an MCP OAuth capable client (claude.ai, Claude Desktop, ChatGPT connectors and similar) you add a connector with no credentials at all, the client discovers the Flashduty authorization server and opens the consent page in your browser, and the token grants the same access as an APP key. Requests with no credential return `401` with a `WWW-Authenticate` header naming the protected-resource metadata. The Go SDK gains the matching `NewClientWithAccessToken`, which authenticates with `Authorization: Bearer` and no longer injects the `app_key` query parameter. See [MCP Server](/en/developer/mcp-server) and [Go SDK](/en/developer/go-sdk).
+
+### CLI: `--timezone` for automation rules
+
+`flashduty automation create` and `flashduty automation update` gain `--timezone` (an IANA zone name): when omitted, `create` lets the server take the caller's member time zone, then the account time zone, then fall back to `Asia/Shanghai`; when omitted on `update` the stored zone is kept, while an explicit empty value is stored as UTC — and the next fire time is recalculated immediately. `alert list` also gains `--alert-ids`, `--alert-keys`, `--asc` and `--by-updated-at`. See [CLI](/en/developer/cli).
+
+</Update>
+
+
 <Update label="2026-09-30" description="📊 Dashboard comparisons and cross-panel cursor linking, RUM apps link to code repositories, on-prem License expiry reminders">
 
 ### Monitors dashboards: comparisons, cross-panel cursor linking and point-value filtering
diff --git a/en/developer/cli.mdx b/en/developer/cli.mdx
index 86aa02833..6c3c6dbc9 100644
--- a/en/developer/cli.mdx
+++ b/en/developer/cli.mdx
@@ -42,6 +42,9 @@ Read the latest version from the [CDN latest version pointer](https://static.fla
 | `FLASHDUTY_INSTALL_DIR` | Custom install directory | `/usr/local/bin` (shell), `~\.flashduty\bin` (PowerShell) |
 | `MIRROR_URL` | Override the release asset download mirror prefix used by the installer script; must use `https://` | `https://static.flashcat.cloud/flashduty-cli` |
 | `FLASHDUTY_UPDATE_BASE_URL` | Override the base URL used by `flashduty update` and the automatic update check | `https://static.flashcat.cloud/flashduty-cli` |
+| `INSTALLED_NAME` | Command name to install as, e.g. `fduty` | `flashduty` |
+
+A copy installed under another command name uses that name in help output, error messages, and shell completion scripts; `flashduty update` replaces the running binary in place, under its own name and directory, instead of installing a default-named copy elsewhere.
 
 ## Authentication
 
@@ -61,6 +64,17 @@ The CLI resolves credentials in the following order (highest priority first):
 2. `FLASHDUTY_APP_KEY` environment variable
 3. Config file `~/.flashduty/config.yaml` (written by `flashduty login`)
 
+### Environment variables
+
+The CLI reads the following environment variables at runtime (variables understood by the installer scripts are listed above under "Installer options"):
+
+| Variable | Description | Default |
+|----------|-------------|---------|
+| `FLASHDUTY_APP_KEY` | APP Key, overriding `app_key` in the config file | - |
+| `FLASHDUTY_BASE_URL` | API endpoint, overriding `base_url` in the config file | `https://api.flashcat.cloud` |
+| `FLASHDUTY_UPDATE_BASE_URL` | Override the base URL used by `flashduty update` and the automatic update check | `https://static.flashcat.cloud/flashduty-cli` |
+| `FLASHDUTY_NO_UPDATE_CHECK` | Set to `1` to disable the daily background update check; the CLI also skips the check in CI environments (`CI`, `GITHUB_ACTIONS`, `JENKINS_URL`, `GITLAB_CI`) | - |
+
 ### Config file
 
 Stored at `~/.flashduty/config.yaml` with `0600` permissions:
@@ -104,6 +118,7 @@ flashduty incident merge <target-id> --source <ids> # Merge incidents into a tar
 flashduty incident timeline <id>         # View an incident timeline
 flashduty incident alerts <id>           # List alerts associated with an incident
 flashduty incident similar <id>          # Find similar historical incidents
+flashduty incident feed <id> [flags]    # View an incident's paginated feed (timeline entries)
 ```
 
 Common filter flags for `incident list`:
@@ -134,6 +149,8 @@ Common filter flags for `incident list`:
 
 Time format examples: `5m`, `1h`, `24h`, `168h`, `2026-04-01`, `2026-04-01 10:00:00`, `1712000000`.
 
+`incident feed <id>` reads an incident's feed (timeline entries) page by page: `--asc` puts the oldest entries first (newest first by default), `--types` filters by comma-separated entry types (e.g. `i_comm,i_assign`; the `type` field of each entry gives the actual type), and `--limit` (max 100) and `--page` paginate the result.
+
 #### Create, update, and merge
 
 `incident create` requires `--title` (3–200 characters) and `--severity`, and prompts for missing values in a terminal. There are three ways to assign people, which can be combined: `--assign` (member IDs), `--assign-emails` (member emails, comma-separated; emails that match no member are ignored), and `--escalate-rule-id` with `--escalate-layer` (assign the people at that zero-based layer of the escalation rule; `--escalate-layer` requires `--escalate-rule-id`).
@@ -202,6 +219,8 @@ flashduty incident comment-type-delete <comment-type-id>      # Delete
 flashduty incident comment-type-reorder <id> [<id2>...]       # Reorder — pass every comment type ID in the account, in the desired order
 ```
 
+To tag a comment with an account-level comment type, pass `--comment-type-id`: `flashduty incident comment <id> [<id2>...] --comment-file <path> --comment-type-id <comment-type-id>`. The comment type ID comes from `flashduty incident comment-type-list`; omit the flag to write the comment without a type.
+
 ### change — Change records
 
 ```bash
@@ -668,6 +687,8 @@ flashduty alert-event list [flags]  # List alert events (default: last 1h)
 
 Common filter flags for both `list` commands: `--severity` (`Critical,Warning,Info`), `--channel` (comma-separated channel IDs), `--integration` (comma-separated integration IDs), `--since`/`--until`, `--limit` (max 100), `--page`. `alert-event list` also takes `--integration-type`, which filters by comma-separated **plugin keys** (e.g. `AliCloud,Prometheus`) — note these are plugin keys, not integration IDs; use `--integration` to filter by integration ID.
 
+`alert list` also takes `--alert-ids` (comma-separated alert IDs), `--alert-keys` (comma-separated alert deduplication keys), `--asc` (sort oldest first by start time; newest first by default), and `--by-updated-at` (apply `--since`/`--until` to the last-updated time instead of the start time). `alert-event list` also supports `--asc` (oldest first; newest first by default), but has no `--by-updated-at` counterpart.
+
 ### automation — AI SRE automation rules
 
 The `automation` command group manages AI SRE automation rules — create, query, update, delete, run history, and triggering. See [Automations](/en/ai-sre/automations) for the console workflow.
@@ -693,18 +714,19 @@ Common `create` flags:
 | `--at` | `HH:MM` local time (in the rule's timezone); for `hourly` only the minute is used (default 0), for `daily`/`weekly` both hour and minute (default 09:00) |
 | `--weekday` | Weekday for `weekly`: `sun`–`sat` or `0`–`7` (`0` and `7` both mean Sunday; defaults to Monday) |
 | `--cron-expr` | Exact 5-field cron expression (interpreted in the rule's timezone); overrides the `--schedule` helpers |
+| `--timezone` | IANA timezone name (e.g. `Asia/Shanghai`) that `--at` and `--cron-expr` are interpreted in; when omitted the server uses the caller's member timezone, then the account timezone, then `Asia/Shanghai`. A timezone the server cannot load is rejected |
 | `--disabled` | Create the rule disabled (enabled by default) |
 | `--schedule-enabled` | Whether the schedule trigger is enabled (default `true`) |
 | `--http-post-trigger` | Create and enable an HTTP POST trigger; when passed alone without any schedule flags, the CLI writes a placeholder cron and disables the schedule trigger |
 | `--prompt` / `--prompt-file` | Task prompt, exactly one required; `--prompt-file` reads from a file (`-` for stdin) |
 | `--environment-kind` / `--environment-id` | Runtime environment: `cloud` or `byoc` (`byoc` requires `--environment-id`); empty means automatic |
 
-`update` accepts `--name`, `--prompt`/`--prompt-file`, `--schedule`/`--at`/`--weekday`/`--cron-expr`, `--enable`/`--disable`, `--enable-schedule`/`--disable-schedule`, `--enable-http-post-trigger`/`--disable-http-post-trigger`, `--rotate-http-post-token`, and `--environment-kind`/`--environment-id`. The personal/team scope is immutable after creation, so `update` exposes no scope flag.
+`update` accepts `--name`, `--prompt`/`--prompt-file`, `--schedule`/`--at`/`--weekday`/`--cron-expr`/`--timezone`, `--enable`/`--disable`, `--enable-schedule`/`--disable-schedule`, `--enable-http-post-trigger`/`--disable-http-post-trigger`, `--rotate-http-post-token`, and `--environment-kind`/`--environment-id`. The personal/team scope is immutable after creation, so `update` exposes no scope flag.
 
 `runs` supports `--status` (`queued`/`running`/`retrying`/`succeeded`/`partial`/`failed`/`skipped`/`abandoned`), `--trigger-kind` (`schedule`/`debug`/`http_post`), `--since`/`--until`, `--page`, and `--limit`. `fire` authenticates with the trigger's token (`--token`, or the `FLASHDUTY_AUTOMATION_TRIGGER_TOKEN` environment variable); `--text` passes context for the run, and `--data` passes a full JSON request body (inline JSON or `-` to read stdin).
 
 <Warning>
-**Timezone semantics**: `--at` and `--cron-expr` are interpreted as local wall-clock time in the **rule's timezone**, which defaults at creation to the caller's member timezone, falling back to the account timezone when the member has none. Pass the user's local time directly — do **not** convert it to UTC. Neither `create` nor `update` has a `--timezone` flag: to pin a different timezone at creation, use the generated command `flashduty safari automation-rule-create --timezone`; a created rule's timezone cannot be changed via `update`.
+**Timezone semantics**: `--at` and `--cron-expr` are interpreted as local wall-clock time in the **rule's timezone**. When `create` omits `--timezone`, the server picks the rule's timezone: the caller's member timezone, then the account timezone, then `Asia/Shanghai`. Pass an IANA name to pin another one (e.g. `--timezone Asia/Shanghai`); a timezone the server cannot load is rejected. When `update` omits `--timezone`, the stored timezone stays, including a legacy empty string; an explicit empty `--timezone ""` is sent and stored as UTC, not the account default. Pass the user's local time directly — do **not** convert it to UTC. `update` recalculates the next fire immediately, so read `schedule_next_fire_at_ms` from the response.
 </Warning>
 
 ### insight — Insight queries
diff --git a/en/developer/go-sdk.mdx b/en/developer/go-sdk.mdx
index e5de5abcf..52cac98b6 100644
--- a/en/developer/go-sdk.mdx
+++ b/en/developer/go-sdk.mdx
@@ -95,7 +95,7 @@ Every call returns three values:
 | `error` | `error` | `*ErrorResponse` on failure, `*RateLimitError` on 429 |
 
 <Tip>
-`app_key` is used for authentication, and the SDK injects it as a query parameter on every request automatically. Obtain the `app_key` from "Push integrations" or team configuration in the Flashduty console.
+`app_key` is used for authentication, and the SDK injects it as a query parameter on every request automatically. Obtain the `app_key` from "Push integrations" or team configuration in the Flashduty console. You can also authenticate with an OAuth access token instead: build the client with `NewClientWithAccessToken` (see [Create a client](#create-a-client)).
 </Tip>
 
 ## Create a client
@@ -132,6 +132,18 @@ client, err := flashduty.NewClient("YOUR_APP_KEY",
 **Private deployment**: point the client at your own Flashduty gateway address with `WithBaseURL` — everything else stays exactly the same.
 </Info>
 
+### OAuth access token
+
+Besides the `app_key`, the SDK also accepts an OAuth access token: a client returned by `NewClientWithAccessToken(accessToken, opts...)` sends `Authorization: Bearer <token>` on JSON, GET, and file-upload requests, and no longer injects the `app_key` query parameter — the two credentials are alternatives, not additive. An empty token returns an error directly (`flashduty: access token is required`); everything else matches `NewClient`, so every `Option` still applies and the default Base URL, timeout, and User-Agent are the same.
+
+```go
+client, err := flashduty.NewClientWithAccessToken(accessToken,
+	flashduty.WithTimeout(10*time.Second),
+)
+```
+
+Access tokens are issued by the Flashduty OAuth authorization server, whose address is the API base URL (default `https://api.flashcat.cloud`). A client discovers it at `/.well-known/oauth-authorization-server` and exchanges an authorization-code + PKCE flow for a token; the token calls the public API as the account it belongs to, with the same access as that account's `app_key`. A token is valid only at the deployment that issued it, so do not point `WithBaseURL` at another gateway when you authenticate with one.
+
 ## Services and methods
 
 ---
diff --git a/en/developer/mcp-server.mdx b/en/developer/mcp-server.mdx
index cba814b03..bff49425f 100644
--- a/en/developer/mcp-server.mdx
+++ b/en/developer/mcp-server.mdx
@@ -38,9 +38,24 @@ Flashduty MCP Server offers three deployment modes; choose based on your MCP cli
 
 <Tabs>
   <Tab title="Remote service (recommended)">
-    No local installation needed — connect directly to the official remote instance `https://mcp.flashcat.cloud/mcp`, authenticating with your APP Key via `Authorization: Bearer`. This is the fastest way to get started.
+    No local installation needed — connect directly to the official remote instance `https://mcp.flashcat.cloud/mcp`; this is the fastest way to get started. The remote service accepts two credential types, based on what your client supports:
 
-    For Cursor, add to your MCP configuration:
+    - **OAuth sign-in**: when your client supports MCP OAuth (such as claude.ai, Claude Desktop, or ChatGPT connectors), add a custom connector with the URL `https://mcp.flashcat.cloud/mcp` and no credentials at all. The client discovers the Flashduty authorization server and opens a consent page in your browser; once you authorize, the client calls the Flashduty Open API as you, with the same access an APP Key grants.
+    - **APP Key**: when your client does not support OAuth, send your APP Key in the `Authorization: Bearer` request header instead.
+
+    OAuth sign-in needs nothing but the URL:
+
+    ```json
+    {
+      "mcpServers": {
+        "flashduty": {
+          "url": "https://mcp.flashcat.cloud/mcp"
+        }
+      }
+    }
+    ```
+
+    With an APP Key, add the following to your MCP configuration (Cursor shown as an example):
 
     ```json
     {
@@ -121,7 +136,8 @@ Flashduty MCP Server offers three deployment modes; choose based on your MCP cli
 | Item | Description |
 | --- | --- |
 | Auth (local) | Provide the Flashduty APP Key via the `FLASHDUTY_APP_KEY` environment variable or the `--app-key` command-line flag. |
-| Auth (remote) | Provide the APP Key via the request header `Authorization: Bearer <your_flashduty_app_key>`. |
+| Auth (remote, OAuth) | Clients that support MCP OAuth need no credential — just authorize Flashduty in the browser. OAuth applies to the remote service only; there is no environment variable or command-line flag for it. |
+| Auth (remote, APP Key) | Provide the APP Key via the request header `Authorization: Bearer <your_flashduty_app_key>`. |
 | Streamable HTTP | `flashduty-mcp-server http`, endpoint `POST /mcp`. The official remote instance is `https://mcp.flashcat.cloud/mcp`. |
 | stdio | `flashduty-mcp-server stdio`, for local clients such as Cursor and Claude Desktop. |
 
@@ -129,6 +145,20 @@ Flashduty MCP Server offers three deployment modes; choose based on your MCP cli
   The server does not support the legacy standalone SSE transport (a separate `GET /sse` endpoint) and returns `405 Method Not Allowed` for `GET` requests. Streamable HTTP already streams responses over `POST` per the MCP spec — use it directly; if your client only supports legacy SSE, upgrade the client or switch to stdio locally.
 </Warning>
 
+### OAuth authorization and discovery
+
+The remote service accepts both MCP OAuth access tokens and APP keys: a credential starting with `oauth:` is treated as an access token, and any other bearer credential is treated as an APP key. Both grant the same access.
+
+A request that carries no credential receives `401` with a `WWW-Authenticate` header naming the protected-resource metadata URL, `Bearer resource_metadata="<request origin>/.well-known/oauth-protected-resource<MCP path>"`. The client uses it to discover the metadata and start the OAuth authorization flow. The server publishes the following metadata endpoints:
+
+| Metadata endpoint | Description |
+| --- | --- |
+| `/.well-known/oauth-protected-resource` | Root path; describes `/mcp`. |
+| `/.well-known/oauth-protected-resource/mcp` | Describes `/mcp`. |
+| `/.well-known/oauth-protected-resource/flashduty` | Describes the legacy `/flashduty` path. |
+
+In each metadata response, `resource` is the request origin plus the matching path, and `authorization_servers` is the Flashduty API address — the address you configure locally with `FLASHDUTY_BASE_URL` or `--base-url` (by default `https://api.flashcat.cloud`), which also acts as the authorization server. An access token is valid only at the API address that issued it, so `base_url` on the remote URL has no effect for OAuth tokens while APP keys still honor it.
+
 ## Configuration
 
 ---
@@ -184,7 +214,7 @@ When building from source and running the binary directly, you can use the follo
 
 ### Remote service configuration
 
-When using the official remote instance, append query parameters to the URL to configure toolsets and read-only mode dynamically:
+When using the official remote instance, append query parameters to the URL to configure toolsets, read-only mode, and the Flashduty API address dynamically:
 
 ```json
 {
@@ -199,6 +229,12 @@ When using the official remote instance, append query parameters to the URL to c
 }
 ```
 
+| Query parameter | Description |
+| --- | --- |
+| `toolsets` | Comma-separated list of enabled toolsets (e.g. `incidents,users,channels`). |
+| `read_only` | Set to `true` to enable read-only mode. |
+| `base_url` | Overrides the Flashduty API address for APP key requests only; an OAuth access token is valid only at the API address that issued it, so it ignores this parameter. |
+
 ### Output format (TOON)
 
 The server supports outputting tool results in [TOON (Token-Oriented Object Notation)](https://github.com/toon-format/toon) format. For arrays of objects with uniform fields (such as member lists or incident lists), TOON can cut token usage by 30%–50% compared with JSON, and modern LLMs can parse it directly. Set `FLASHDUTY_OUTPUT_FORMAT=toon` (or `--output-format toon`) to enable it.
diff --git a/en/monitors/dashboards/dashboards.mdx b/en/monitors/dashboards/dashboards.mdx
index 1672c3242..a87d62cf3 100644
--- a/en/monitors/dashboards/dashboards.mdx
+++ b/en/monitors/dashboards/dashboards.mdx
@@ -96,7 +96,7 @@ The **query type** follows from the data source and the visualization and is not
 
 | Setting | Description |
 |---------|-------------|
-| **Query expression** | Supports `${variable}` templates; PromQL data sources prompt for a PromQL query, other types prompt by type |
+| **Query expression** | Prometheus and Loki panels support `${variable}` templates (the braces are optional — `$variable` works too) as well as built-in time variables such as `$__interval`, see [Built-in time variables](#built-in-time-variables); MySQL panels support no variable interpolation; PromQL data sources prompt for a PromQL query, other types prompt by type |
 | **Multiple queries** | Queries are labelled `A`~`Z`; only time series support more than one. Switching to any other visualization forces you to pick which one to keep, and the rest are removed |
 | **Series name** | Shown for time series only. Sets the display name used by the legend, tooltip and data links; `{{instance}}` is replaced with that series' `instance` label value, and leaving it empty shows all labels. If both the legend and the tooltip are hidden, the editor tells you the setting is invisible |
 | **Min step** | Shown for Prometheus range queries only. A lower bound for the query step, 15s by default (matching Prometheus' default scrape interval); it is dropped when you switch to a data source or query type that does not accept it |
@@ -121,6 +121,29 @@ Durations are always written as a number plus a unit (`s` / `m` / `h` / `d` / `w
 
 Previewing happens live inside the panel editor: editing the query itself runs it on Enter (or `⌘/Ctrl + Enter`), while changing the datasource, visualization or time re-runs it automatically; display settings (unit, thresholds, alias and so on) apply immediately without re-querying. **Saving does not require a successful preview.** On a new dashboard with no folder selected yet, the preview area shows "Select a folder before previewing".
 
+### Built-in time variables
+
+A panel query expression can reference built-in time variables directly; they render the seconds computed from that panel's own execution window and step, so `$__range` spares you from maintaining a hand-written variable for how long the window is.
+
+| Variable | What it renders |
+|----------|-----------------|
+| `$__interval` | The panel's query step in seconds, such as `30s`; the step follows from the time range and the **Max data points**, bounded by the **Min step** |
+| `$__rate_interval` | The larger of `step + 15s` and `60s` (15s is the assumed scrape interval) — the value to hand straight to `rate()` |
+| `$__range` | The length of the panel's execution window in seconds, e.g. `86400s` for a 24-hour window |
+| `$__auto` | In a Loki instant query, the duration of the whole window in the largest unit that divides it — `1h`, `5m`, `90s` or `90500ms`; in a Loki range query it is the step in seconds, like `$__interval` |
+
+The reference syntax matches variables: both `${__interval}` and `$__interval` are accepted (the braces are optional), but a built-in takes no formatter — `${__interval:raw}` and the like are rejected.
+
+Which ones you can use follows from the datasource and the query type; every combination outside the table below (MySQL included) is unavailable, and such a definition fails validation on save — the dialog lists the offending JSON Pointer and the reason — while the panel itself queries nothing at run time:
+
+| Datasource · query type | Available built-in time variables |
+|-------------------------|-----------------------------------|
+| Prometheus · range | `$__interval`, `$__rate_interval`, `$__range` |
+| Loki · range | `$__auto`, `$__interval`, `$__range` |
+| Loki · instant | `$__auto`, `$__range` |
+
+Built-in time variables exist only in a **panel query expression**: neither a template variable's candidate query nor a panel query's `args` can resolve them, and the definition is rejected on save.
+
 ## Template variables
 
 The variable bar sits above the canvas and uses the same filter control as the rest of the product: each variable is a permanent filter field, single- or multi-select according to its configuration. In edit mode the bar has a **Manage variables** button at the end, opening a drawer where every variable can be added, edited or deleted.
@@ -133,9 +156,9 @@ There are three **variable kinds**:
 | **Custom** | A hand-written candidate list | Options (one per line, `text=value` or a bare value) |
 | **Query** | Candidates fetched by querying a data source | Datasource (fixed or a datasource variable) + query definition + when to refresh candidates |
 
-A **Query variable's candidate query** has three flavours by data source type: Prometheus needs `label` (required) and `metric` (optional) plus any number of **label filters** using `= / != / =~ / !~`; SQL data sources take an expression and a value field; Loki takes an expression and a field. **Preview candidates** in the drawer fetches candidates once with the current configuration to confirm the datasource and query are right, and shows the failure reason next to the variable when they are not.
+A **Query variable's candidate query** has three flavours by data source type: Prometheus needs `label` (required) and `metric` (optional) plus any number of **label filters** using `= / != / =~ / !~`; SQL data sources take an expression and a value field; Loki takes an expression and a field, where the field must be a stream label name (`^[A-Za-z_][A-Za-z0-9_]*$` — dotted or JSON field names do not work), and the candidates are that label's values on the streams the expression selects **within the dashboard's time range**, with an empty expression meaning every stream; the expression may reference dashboard variables but not built-in time variables (those exist only in panel query expressions, see [Built-in time variables](#built-in-time-variables)). **Preview candidates** in the drawer fetches candidates once with the current configuration to confirm the datasource and query are right, and shows the failure reason next to the variable when they are not.
 
-**Names and references**: a variable name must match `^[A-Za-z_][A-Za-z0-9_]{0,63}$`, and a name already in use is rejected. Renaming is an atomic frontend refactor — every `${oldName}` reference (query expressions, query args, datasource variable references and the variable's own candidate query) is rewritten, and the editor warns "N references will be rewritten" before doing it. A variable that is still referenced cannot be deleted: "Variable x is referenced in N places; remove the references first." A dashboard holds at most 20 variables.
+**Names and references**: a variable name must match `^[A-Za-z_][A-Za-z0-9_]{0,63}$` and must not start with `__` (that prefix is reserved for built-in time variables), and a name already in use is rejected. Renaming is an atomic frontend refactor — every `${oldName}` reference (query expressions, query args, datasource variable references and the variable's own candidate query) is rewritten, and the editor warns "N references will be rewritten" before doing it. A variable that is still referenced cannot be deleted: "Variable x is referenced in N places; remove the references first." A dashboard holds at most 20 variables.
 
 **Selection mode and the 「All」 option**:
 
@@ -151,6 +174,15 @@ The URL is the single source of truth for variable selections: a single value is
 
 - The **time range** comes from the URL's `from` / `to`, either absolute milliseconds or a `now-*` relative expression; with no URL parameters the stored **default time range** applies (the last hour for a new dashboard, and one of eight presets — 15m / 30m / 1h / 3h / 6h / 12h / 24h / 7d — when saved, always ending at "now").
 - **Auto-refresh**: the interval stored on the dashboard is one of Off / 30s / 1m / 5m; the viewer's refresh picker offers finer steps (Off / 10s / 30s / 1m / 5m / 15m / 30m / 1h). Picking one writes the `refresh` URL parameter and refreshes immediately rather than waiting a full interval. Relative ranges roll forward with "now" on each refresh, and the refresh button on the left runs a manual refresh at any time, spinning while queries are still running.
+- **Display time zone**: the bottom of the time picker's dialog holds a time zone row — the current zone (city plus abbreviation) and its UTC offset on the left, and a **Change time zone** button on the right that opens a searchable city list (search by city name, abbreviation or IANA name), grouped by continent, whose first group **Common** holds the three special entries in the table below. It applies to dashboards only (viewer, edit mode, panel editor and panel fullscreen) and decides how the time picker displays and parses times, the time series x-axis and tooltip times, and the time range label; what actually renders in the zone is absolute instants, while a relative range (such as the last hour) is a duration and does not shift with the zone. When the display zone differs from your personal effective zone, the time picker carries that zone's abbreviation as a reminder.
+
+| Common entry | Meaning |
+|--------------|---------|
+| **Default** | Follow your personal effective time zone |
+| **Browser time** | Follow the current browser's time zone |
+| **Coordinated Universal Time** | Always UTC |
+
+The choice is stored in this browser (the localStorage key `monit.dashboards.timeZone`) and reused when you come back; picking **Default** clears that record and returns to your personal effective zone.
 
 ## Period-over-period comparison
 
diff --git a/en/monitors/explore/explore.mdx b/en/monitors/explore/explore.mdx
index fa21f76a4..49d172a74 100644
--- a/en/monitors/explore/explore.mdx
+++ b/en/monitors/explore/explore.mdx
@@ -78,7 +78,32 @@ In addition to the query expression, some data sources support private parameter
 | Project / Logstore | SLS | Required; used in the request URL |
 | Region / Topic / Syntax | Tencent CLS | Required |
 | PowerSQL | SLS | Enables PowerSQL mode |
-| Time macros | SQL data sources | `$__timeFilter(column)`, `$__timeFrom()`, `$__timeTo()` make the time range take effect |
+| Time macros | SQL data sources | `$__timeFilter(column)`, `$__timeFrom()`, `$__timeTo()`; the time range you pick on the page reaches this query only when the SQL uses a macro — without one the SQL runs as written |
+| Built-in time variables | Prometheus / Loki | `$__interval`, `$__rate_interval`, `$__range`, `$__auto` make the query follow the time range and step you pick on the page — see the next section |
+
+### Built-in Time Variables
+
+Prometheus and Loki queries support built-in time variables: before the query runs, each variable is rendered into a concrete duration from that query's own time window and step, so the query follows the time range you select on the page without hard-coding any time. Write a variable with a leading `$`, or in the `${...}` form.
+
+| Variable | Rendered value |
+|----------|----------------|
+| `$__interval` | The step of this query in seconds, for example `30s`; it is the same step that is sent to Edge — the larger of the value the backend derives from the data point count and **Min step** |
+| `$__rate_interval` | The step plus 15 seconds, and never below 60 seconds, for example `60s`; 15 seconds is the scrape interval `$__rate_interval` assumes |
+| `$__range` | The length of this query's time window in seconds, with a minimum of 1 second, for example `3600s` |
+| `$__auto` | In a Loki range query, the step (for example `30s`); in a Loki instant query it would be the whole time window in the largest unit that divides it (for example `1h`), but instant queries carry no start time, so it is not usable (see below) |
+
+Availability depends on both the data source and the query type:
+
+| Query | Available built-in time variables |
+|-------|----------------------------------|
+| Prometheus range (**Graph only**, and the graph half of **Graph + Raw**) | `$__interval`, `$__rate_interval`, `$__range` |
+| Prometheus instant (**Raw only**, and the raw half of **Graph + Raw**) | none |
+| Loki range | `$__auto`, `$__interval`, `$__range` |
+| Loki instant | none |
+
+- No built-in time variable works in an instant query: a Prometheus instant query accepts no range variable, and a Loki instant query carries no start time, so window-length variables such as `$__auto` and `$__range` cannot be rendered. If you write one, the result area shows "The query or its parameters are invalid. Check them and retry." with the backend reason below it (for example `built-in variable "__rate_interval" is not available in prometheus instant queries`, `built-in variable "__range" requires from_ms`).
+- Prometheus's default **Graph + Raw** query type sends one range query and one instant query that share the same statement: `$__rate_interval` renders on the graph half and fails on the raw half. To use `$__rate_interval` in this query type, switch the query type to **Graph only**.
+- Built-in time variables share their names with the dashboard built-in time variables and are computed by the same rendering logic from this query's time window and step. Apart from the variable names in the table above, any other `$` text — including `${name}` and regex capture-group references such as `$1` — is sent to the data source verbatim and is not substituted, because the workbench has no dashboard variables.
 
 ## AI Query Assistant
 
diff --git a/en/openapi/go-sdk.mdx b/en/openapi/go-sdk.mdx
index 1829ddb6b..ee32029d0 100644
--- a/en/openapi/go-sdk.mdx
+++ b/en/openapi/go-sdk.mdx
@@ -10,7 +10,7 @@ keywords: ["Go SDK", "go-flashduty", "Golang", "Open API", "client", "SDK"]
 If your service is written in Go, use the SDK to call Flashduty rather than hand-writing HTTP requests — you get typed request parameters, response structs, and error codes out of the box, with no manual JSON assembly or response parsing.
 
 <Tip>
-The SDK shares the same endpoint semantics as the Open API. Every request/response structure documented here maps to a corresponding type in the SDK. Authentication uses the same [APP Key](/en/openapi/introduction#authentication).
+The SDK shares the same endpoint semantics as the Open API. Every request/response structure documented here maps to a corresponding type in the SDK. Authentication uses the same [APP Key](/en/openapi/introduction#authentication), or an OAuth access token — see [Options](#options).
 </Tip>
 
 ## Install
@@ -91,6 +91,16 @@ client, err := flashduty.NewClient("YOUR_APP_KEY",
 )
 ```
 
+### OAuth access token
+
+Besides `NewClient`, the SDK exports `NewClientWithAccessToken(accessToken, opts...)`: it authenticates with an OAuth access token, sending `Authorization: Bearer <token>` on JSON, GET, and file-upload requests, and does not inject the `app_key` query parameter; an empty token returns an error directly, and the options are exactly the same as for `NewClient`. The token is issued by the Flashduty OAuth authorization server (the API base URL) and is valid only at the deployment that issued it.
+
+```go
+client, err := flashduty.NewClientWithAccessToken(accessToken,
+	flashduty.WithTimeout(10*time.Second),
+)
+```
+
 ## Errors and rate limits
 
 The SDK returns typed errors you can unwrap with `errors.As`, plus convenience predicates that see through wrapped errors:
diff --git a/en/rum/explorer/analytics.mdx b/en/rum/explorer/analytics.mdx
new file mode 100644
index 000000000..5a5dabea4
--- /dev/null
+++ b/en/rum/explorer/analytics.mdx
@@ -0,0 +1,139 @@
+---
+title: "RUM Explorer Analytics"
+sidebarTitle: "Analytics"
+description: "Summarize RUM events in the Flashduty RUM Explorer analytics view with aggregations, groups and five chart types, then drill from an aggregate down to the raw events."
+keywords: ["RUM", "data explorer", "analytics", "charts", "drill-down"]
+---
+
+The RUM Explorer gives you two view modes: **List** and **Analytics**. The **Analytics** view aggregates the events of the current event type by the aggregation and group you pick, which is what answers questions like "which page loads slowest" or "which browsers do the errors come from". Switch back to the list when you need individual events.
+
+## List and analytics
+
+The top of the Explorer has two tabs: **List** shows events one by one, **Analytics** aggregates them into charts. Both share the same event type, query conditions and time range, so switching tabs never drops your current filters.
+
+The mode is recorded in the `mode` URL parameter: `mode=analytics` opens analytics, and omitting `mode` opens the list, so existing Explorer links keep opening the list. Copy the URL of an analytics view to share the same chart with others.
+
+| Tab | `mode` value | What it shows |
+|-----|--------------|---------------|
+| **List** | omitted | Event trend graph and event table, drilling into individual event details |
+| **Analytics** | `analytics` | Charts aggregated by the aggregation and group |
+
+## Choosing the aggregation and fields
+
+The **Measure** row at the top of the analytics view decides what is computed: pick an **Aggregation**, then a **Measured field** where one is needed, then a **Group-by field**.
+
+### Aggregations
+
+| Aggregation | `agg` value | Description |
+|-------------|-------------|-------------|
+| **Count** | `count` | Number of events; takes no field |
+| **Unique count** | `count_distinct` | Number of distinct values of a field |
+| **Average** | `avg` | Mean of a numeric field |
+| **Sum** | `sum` | Total of a numeric field |
+| **Min** | `min` | Smallest value of a numeric field |
+| **Max** | `max` | Largest value of a numeric field |
+| **P50 / P75 / P90 / P95 / P99** | `p50` / `p75` / `p90` / `p95` / `p99` | Approximate percentiles of a numeric field, for watching the long tail |
+
+Every aggregation except **Count** needs a **Measured field**. With Average, Sum, Min, Max or any percentile, the field picker lists numeric fields only; timestamp fields (unit `unix_millisecond`) are excluded, since averaging them is meaningless. With **Unique count**, the field picker lists the scalar facets that can be grouped by. After you switch the event type, a field that does not exist on the new event type is reported as missing and you are asked to pick another one, rather than the question silently becoming a count.
+
+The legend, the table columns and the CSV header all use the same naming: the measure of **Count** and of a **Distribution** reads "Events"; every other aggregation reads "Aggregation (field name)", for example "Average (view_loading_time)".
+
+### Group-by field
+
+The **Group-by field** decides what the measure is split by. Once one is chosen, the chart draws each group value separately and a **top** selector appears next to it, offering `5`, `10`, `20` and `50` groups, `10` by default. The group-by field can be cleared (the placeholder reads "No grouping"), and the result then collapses into a single number. Only scalar facets can be grouped by; an array field holds several values per event and therefore has no meaningful grouping.
+
+A **Distribution** does not use a group-by field, and that selector is not shown.
+
+## Chart types
+
+The analytics view offers five chart types:
+
+| Chart type | `viz` value | Description |
+|------------|-------------|-------------|
+| **Timeseries** | `timeseries` | The measure over time buckets; the default chart type |
+| **Top list** | `toplist` | Groups ranked by the measure, largest first |
+| **Pie** | `pie` | Share of each group, subject to the constraints below |
+| **Table** | `table` | Two columns: the group and the measure |
+| **Distribution** | `distribution` | Histogram of events per value range of one numeric field |
+
+A **Timeseries** also offers a **Draw style** selector that switches between **Line** (`line`) and **Bars** (`bar`), bars by default. When a group-by field is set, the draw style is bars and the aggregation is **Count** or **Sum**, the bars are stacked by group and the bar height is the total; the other aggregations (averages, percentiles, unique counts) cannot stack, because their per-group values do not add up.
+
+Without a group-by field, **Top list** and **Table** print the measure itself instead of drawing a single bar; a **Timeseries** still plots one line over time.
+
+## Drilling down
+
+Clicking an element on a chart opens a drill-down menu that takes you from the aggregate back to individual events: a point in time on a timeseries, a row in the top list or table, a slice of a pie, or a range of a distribution.
+
+| Menu item | What it does |
+|-----------|--------------|
+| **View in list** | Switches to the list tab, showing only the events behind the clicked element |
+| **Split "value"** | Keeps only the clicked group value and opens the group-by picker so you can break it down by another field; a distribution range has no group, so the item reads **Filter to "range"** and keeps only the events in that range |
+| **Exclude "value"** | Drops the clicked group value from the result |
+
+A drill-down writes the same filters the facet list on the left writes, so they show up in the search box and can be undone the same way. Drilling down from a timeseries point narrows the list to that time bucket, and drilling down from a distribution range filters the list by that numeric range.
+
+After **View in list**, a **Back to analytics** bar appears above the list with the note "The list is filtered by the drill-down; going back restores the previous query". Clicking it returns to the analytics view and restores the query you drilled from. Switching tabs by hand ends that drill-down, and the bar disappears with it.
+
+<Note>
+The empty group, shown on charts as "(empty)", cannot be drilled into yet, so **Split** and **Exclude** are both unavailable for it.
+</Note>
+
+## Zooming a timeseries
+
+Drag across a **Timeseries** to zoom into that window. The zoom applies to this chart only; the Explorer's own time range stays as it is. While zoomed, the **Zoomed** badge in the top-right corner of the chart shows the current window, and the close button on it (**Reset zoom**) restores the full time range. Changing the event type, the query conditions or the Explorer's time range resets the zoom.
+
+## Downloading an image and exporting CSV
+
+Two export buttons sit in the top-right corner of the chart:
+
+| Button | Description |
+|--------|-------------|
+| **Download image** | Exports the current chart as a PNG rendered at twice the pixel density. A Table, or a single number without a group-by, is not a chart, so the button is unavailable |
+| **CSV** | Exports the current chart's data as CSV, named like `rum-<event type>-analytics-<timestamp>.csv`. A timeseries exports one row per time, group and measure (long format); the other charts export one row per group |
+
+Both buttons are unavailable while the result is empty.
+
+## Distribution
+
+A **Distribution** buckets one numeric field by value range and shows the number of events in each range. It ignores the aggregation and the group-by.
+
+- Buckets span from the field's minimum up to `P99`: RUM timings have a long tail (one 46s page load among thousands of 1s ones), and splitting up to the maximum would crush the whole body of values into the first bar.
+- The bucket width is rounded to `1`, `2` or `5` times a power of ten, so bucket edges read as round numbers; the observed range is split into about 40 buckets.
+- Everything above `P99` goes into one open-ended last bucket, shown on the chart and in the CSV as "≥ edge".
+- Ranges are labelled in the field's own unit, for example `1s–1.5s` for a duration field.
+
+## When a Pie applies
+
+A Pie requires both of the following; otherwise the chart area shows "Pie charts need a group-by and only apply to count and sum":
+
+- a **Group-by field** is set;
+- the aggregation is **Count** or **Sum** — only these two add up to a whole, which is what a pie claims. Unique counts, averages and percentiles do not.
+
+## Sampled estimates
+
+When the data volume is large enough that the result is estimated from a sample, the **Sampled** badge appears in the top-right corner of the chart, with the tooltip "Large data volume: results are estimated from a sample".
+
+## URL parameters
+
+The analytics view keeps its state in the URL, so a copied link reproduces the same analysis. Values that are not allowed fall back to the default.
+
+| Parameter | Meaning | Values | Default |
+|-----------|---------|--------|---------|
+| `mode` | View mode | `analytics` | omitted, i.e. the list |
+| `agg` | Aggregation | `count`, `count_distinct`, `avg`, `sum`, `min`, `max`, `p50`, `p75`, `p90`, `p95`, `p99` | `count` |
+| `aggField` | Measured field | a field name | none |
+| `groupBy` | Group-by field | a field name | none (no grouping) |
+| `top` | Number of groups kept | `5`, `10`, `20`, `50` | `10` |
+| `viz` | Chart type | `timeseries`, `toplist`, `pie`, `table`, `distribution` | `timeseries` |
+| `style` | Timeseries draw style | `line`, `bar` | `bar` |
+
+## Next steps
+
+<CardGroup cols={2}>
+  <Card title="RUM Explorer Overview" icon="compass" href="./overview">
+    Learn about the Explorer's core features
+  </Card>
+  <Card title="Data Query Guide" icon="magnifying-glass" href="./data-query">
+    Master query syntax for precise data location
+  </Card>
+</CardGroup>
diff --git a/en/rum/explorer/overview.mdx b/en/rum/explorer/overview.mdx
index 2422c00c3..96e8d236d 100644
--- a/en/rum/explorer/overview.mdx
+++ b/en/rum/explorer/overview.mdx
@@ -21,6 +21,9 @@ Flashduty RUM **Explorer** is a powerful data analysis tool designed to help dev
   <Card title="Data Search & Filtering" icon="filter">
     Precisely filter and select RUM events through the search bar and visualization type selection
   </Card>
+  <Card title="Analytics" icon="chart-line" href="./analytics">
+    Aggregate RUM events into charts by aggregation and group, and drill from an aggregate down to the raw events
+  </Card>
 </CardGroup>
 
 ## Value and Benefits
@@ -148,6 +151,9 @@ The event table in the Sessions Explorer provides a **CSV Export** button that e
   <Card title="Data Query Guide" icon="magnifying-glass" href="./data-query">
     Master query syntax for precise data location
   </Card>
+  <Card title="Analytics" icon="chart-line" href="./analytics">
+    Summarize RUM events with aggregations, groups and charts
+  </Card>
   <Card title="Distributed Tracing" icon="diagram-project" href="../best-practices/distributed-tracing">
     Learn distributed tracing best practices
   </Card>
diff --git a/en/rum/quickstart/app-management.mdx b/en/rum/quickstart/app-management.mdx
index 9ac0c1d4a..bc30e9c8b 100644
--- a/en/rum/quickstart/app-management.mdx
+++ b/en/rum/quickstart/app-management.mdx
@@ -82,6 +82,8 @@ The console provides detailed integration guides for each platform:
 
 Each platform's SDK configuration page automatically fills in the current application's `applicationId` and `clientToken`, so you can copy the code directly into your project.
 
+The Web, Android, iOS, Flutter, HarmonyOS, and WeChat Mini Program SDK configuration pages all provide an **Allow remote configuration** switch, which is on by default. When it is on, the generated initialization code carries `remoteConfigurationEnabled: true` (on Android and HarmonyOS, the equivalent `.setRemoteConfigurationEnabled(true)` initialization parameter); when it is off, that line does not appear in the code and the client does not request the remote configuration at all. The switch only affects the generated code — you still have to enable and publish a configuration on the [Remote Configuration](#remote-configuration) tab, and until you publish, the client keeps using the configuration in your code.
+
 <Warning>
 Modifying SDK configuration in application management will not take effect in real-time on integrated clients. All configuration changes need to be updated in your frontend code and redeployed to take effect.
 </Warning>
@@ -132,6 +134,7 @@ When the application type is **WeChat Mini Program**, **Application Configuratio
 | `service` | Service name; all events default to this tag | Letters, digits, underscore only; up to 24 chars | — |
 | `version` | Version number, useful for filtering during analysis | Letters, digits, `.` only; up to 24 chars | `1.0.0` |
 | `sessionSampleRate` | Session sample rate (percentage) | Integer in range `0–100` | `10` |
+| `remoteConfigurationEnabled` | Whether the generated initialization snippet enables remote configuration | — | On |
 
 <Note>
 No save action is needed — the preview snippet updates as you type. `applicationId` and `clientToken` are filled in automatically by the system; you do not need to set them by hand.
@@ -160,7 +163,8 @@ flashcatRum.init({
   service: "<SERVICE_NAME>",
   env: "<ENV_NAME>",
   version: "1.0.0",
-  sessionSampleRate: 10
+  sessionSampleRate: 10,
+  remoteConfigurationEnabled: true // Apply settings published in the console
 });
 ```
 </Step>
@@ -390,7 +394,7 @@ For reference, the protocol semantics: besides the default values, a configurati
 
 - Change history keeps the most recent 50 versions; older versions are cleaned up automatically and cannot be rolled back to. Each entry shows who made the change and the change note (if one was left). You can revert any retained version to make it the current configuration; a revert also produces a new version.
 - A new version produced by a revert carries a content-equivalence label: if its configuration content is identical to an earlier version in the history, the history list marks it as equivalent to the earliest version with that content — a version-number jump does not mean the configuration content changed.
-- The rollout panel shows **adoption** and rollout progress across recently arrived sampled sessions, with status labels *waiting for new sessions* (just published, no session has started under the current version yet), *in effect* (some sessions have switched), *recent sessions switched* (every session that arrived in the window is on the current version), *no configuration received* (none of the recent sessions reported a configuration version — usually the SDK has remote configuration disabled or is too old), *no data* (no new sessions in the window), and *no adoption detected* (published for over about 8 hours and still not on the current version); slices that did not report a configuration version are labeled *not reported* (SDK too old or config never fetched). Clicking **View details** opens the event explorer on this batch of sampled sessions. When fewer than 20 sampled sessions fall in the window, the panel shows the actual session counts only and does not estimate shares (shares are extrapolated from the sample rate, which is meaningless with too few samples).
+- The rollout panel shows **adoption** across recently arrived sampled sessions: a waffle chart of 10×10 = 100 squares represents each version's share of sessions, where each square is 1% of sessions, estimated from each version's sample rate, and the detail table next to it lists the **Version**, **Share**, and **Sessions** of every group. The figures above show four statistics — **Active sessions**, **Reported**, **Not reported**, and **Current version adoption** — and current-version adoption carries a check mark labelled *recent sessions switched* once every session that arrived in the window has switched. Groups are merged as follows: the current version forms a group of its own, labelled *Current*; the two older versions with the highest share each form their own group; the remaining older versions are merged into *Earlier versions*; and sessions that did not report a configuration version are listed separately as *Not reported* (SDK too old or config never fetched). Status labels are *waiting for new sessions* (just published, no session has started under the current version yet), *in effect* (some sessions have switched), *recent sessions switched* (every session that arrived in the window is on the current version), *no configuration received* (none of the recent sessions reported a configuration version — usually the SDK has remote configuration disabled or is too old), *no data* (no new sessions in the window), and *no adoption detected* (published for over about 8 hours and still not on the current version). Clicking **View session details** opens the event explorer on this batch of sampled sessions; on the version card, **View config** expands the full content of the current configuration, and **Hide config** collapses it again. When fewer than 20 sampled sessions fall in the window, the panel draws no waffle chart and shows the actual session counts per group only, without estimating shares (shares are extrapolated from the sample rate, which is meaningless with too few samples).
 
 ## Delete Application
 
diff --git a/zh/ai-sre/automations.mdx b/zh/ai-sre/automations.mdx
index 85ac8ddea..b7e9a6f9d 100644
--- a/zh/ai-sre/automations.mdx
+++ b/zh/ai-sre/automations.mdx
@@ -58,6 +58,7 @@ sidebarTitle: 自动化
 | 范围 | 是 | 通过 **范围选择器** 选 **个人**（`team_id=0`）或某个 **团队**（`team_id>0`）。范围既决定这条规则的归属与编辑权限，也限定 **执行 Environment** 里可选的自托管 Runner——只有账户全局的 Runner，以及与该范围同团队的 Runner 才可选。 |
 | 执行环境 | 否 | 通过 **环境选择器** 选运行环境：**自动**（由后端挑选最优可用环境，默认值）、**云端沙箱**，或某个 **自托管（BYOC）Runner**。选了某个团队范围后，不属于该范围的团队 Runner 会被自动清除。 |
 | 任务提示词 | 是 | 描述要让 AI SRE 执行的任务，用富文本编辑器撰写。这段提示词就是每次运行时发给 Agent 的内容。占位提示：`描述 Flashduty AI SRE 要执行的任务。` |
+| 时区 | 否 | **按周期执行** 触发卡片里的字段（见下文「按周期执行（cron）」）：cron 表达式所依据的 **规则时区**，用一个可搜索的下拉框选择 IANA 时区名（如 `Asia/Shanghai`），选项按下拉框里的 UTC 偏移排列。新建规则时默认为你的生效时区（成员时区优先，成员未设置时回退账户时区）；没有可用时区时该字段留空，创建请求不携带它。 |
 
 <Note>
 **执行环境** 的「自动」会在每次运行时由后端挑选当前最优的可用环境；「云端沙箱」是平台托管的临时沙箱；自托管 Runner 则把运行落在你自己的机器上。三者的差异与连接方式见 <a href="/zh/ai-sre/environments">运行环境</a>。
@@ -96,13 +97,17 @@ sidebarTitle: 自动化
 | 自定义 | 直接填 4 段或 5 段 cron 表达式 |
 
 <Warning>
-**时区**：在 **每天 / 每周** 模式下，你选的时刻按 **本地时区** 理解，保存时会换算成 UTC，界面会在节奏摘要旁标注你的本地时区；**自定义** 模式下表达式按 **UTC** 解释，界面标注为 `UTC`。
+**时区**：**按周期执行** 的运行按 cron 的 **规则时区** 读表——时区在触发卡片里的 **时区** 下拉框中选定，**每小时 / 每天 / 每周 / 自定义** 四种模式共用这一个字段，不存在「哪种模式按本地时区、哪种模式按 UTC」的区别。保存时按你选中的 IANA 时区原样存入规则，不再把时分换算成 UTC，节奏摘要旁也不再另外标注时区。
 </Warning>
 
 <Note>
 实际执行时间可能与设定时间存在 **分钟级延迟**，这是有意为之，用于把系统负载分散开。请不要把规则当作秒级精确的定时器使用。
 </Note>
 
+编辑一条 **已启用**、且已经排好下一次运行（存在待执行的那一次）的规则时，改动 cron 表达式或 **时区** 会在触发卡片里多出一条提示：**已经排好、还没开始的那一次（`<时刻>`，`<原时区>`）不会再跑。下一次按 `<新时区>` 的钟面重算，实际开始最多比那个分钟再晚 5 分钟。**
+
+只有确实改到了排期才会出现这条提示：把时区在空值与 `UTC` 之间切换不算改动（两者是同一个钟面），关闭 **按周期执行** 触发器、或这条规则当前没有待执行的运行时也不会出现。
+
 ### 经 API 调用（HTTP POST）
 
 让你在外部系统里按需触发这条自动化，而不依赖时间周期。
diff --git a/zh/ai-sre/im.mdx b/zh/ai-sre/im.mdx
index 12701579f..6dd669657 100644
--- a/zh/ai-sre/im.mdx
+++ b/zh/ai-sre/im.mdx
@@ -83,13 +83,13 @@ IM 交互依赖你已在 Flashduty 中接入对应平台的机器人（用于告
 
 ### 处理中回执
 
-收到你的消息后，机器人会先给**你发出的那条消息**贴一个表情回执，作为「已收到、正在处理」的即时信号。表情从该平台自己的表情池里**随机**挑选，不再固定某一个：
+收到你的消息后，机器人会先给**你发出的那条消息**贴一个表情回执，作为「已收到、正在处理」的即时信号。表情从该平台自己的表情池里**随机**挑选，不再固定某一个——池子里只保留表示「已收到、正在处理」的字形，道贺、鼓掌、点赞、比心、笑脸这类会被读成结论或反话的表情不在其中：
 
 | 平台 | 表情池（随机取一个） |
 | --- | --- |
 | Slack | eyes、thinking_face、hourglass_flowing_sand、mag、zap |
-| 飞书 / Lark | Typing、Thinking、Get、CheckMark、OnIt、Hundred 等 |
-| 钉钉 / DingTalk | 平台内置表情：思考、收到、Get、赞、比心、OK 等 |
+| 飞书 / Lark | Typing、THINKING、Get、OnIt |
+| 钉钉 / DingTalk | 平台内置表情：思考、收到、Get |
 
 Slack 除此之外还会显示一条原生的「正在处理」状态，与表情回执并存。
 
diff --git a/zh/ai-sre/mcp.mdx b/zh/ai-sre/mcp.mdx
index 800610547..bf7233db3 100644
--- a/zh/ai-sre/mcp.mdx
+++ b/zh/ai-sre/mcp.mdx
@@ -55,7 +55,7 @@ sidebarTitle: MCP
     对话框先展示模板信息：供应商（如「Datadog 官方 · MCP」或「社区版 · MCP」）、官方文档链接、简介，以及一组「**连接后可以这样问**」示例问题。随后是模板声明的连接参数字段——URL / 端点占位符与环境变量等；**共享**字段（服务授权模式下全账号共用）下方标注「全账号共用。」。若模板是「成员授权（密钥）」模式，另有「**你的 API Token**」字段（密码输入框），下方标注「只给你自己用，其他成员第一次使用时各自填写。」，并在字段旁给出「如何获取」文档链接。
   </Step>
   <Step title="选择运行环境并提交">
-    在「**运行环境**」里选择这次连接从哪个环境发起——默认在会话所在的环境里连接；模板需要自托管 Runner 时，提示「〈名称〉只在内网能访问时，选内网里的自托管 Runner。」。主按钮是 **连接**；若模板是「成员授权（OAuth）」模式，主按钮为「**去〈名称〉授权**」，点击后先创建连接器、再打开浏览器授权窗口（OAuth 的发现、动态客户端注册与令牌交换都从所选环境发起）。
+    在「**运行环境**」里选择这次连接从哪个环境发起——默认在会话所在的环境里连接；模板需要自托管 Runner 时，提示「〈名称〉只在内网能访问时，选内网里的自托管 Runner。」。主按钮是 **连接**；若模板是「成员授权（OAuth）」模式，主按钮为「**去〈名称〉授权**」，点击后先创建连接器、再打开浏览器授权窗口（OAuth 的发现、动态客户端注册与令牌交换都从所选环境发起；授权服务器不支持 DCR 时改用 Client ID Metadata Document，见下文「认证」）。
   </Step>
   <Step title="确认连接结果">
     提交后系统创建连接器并对它执行一次连接测试。通过则显示「**〈名称〉已连接**」与示例问题，点 **完成** 关闭对话框，或直接点某个示例问题新建一条**已预填**该问题的对话（不会自动发送）。不通过则显示「**连接测试没通过**」与原始错误：可点 **重试** 重跑测试；需要 Runner 的模板在非云端环境下还会给出「让 AI SRE 在 Runner 上装好」入口，新建一条不含任何令牌的对话，让 Agent 到自托管 Runner 上安装缺失的命令行工具（如 `npx` 对应的 Node.js、`uvx` 对应的 uv）。
@@ -158,7 +158,7 @@ MCP 服务器支持三种**授权模式**，决定凭证如何提供给服务器
     适合每个工程师使用各自 API Key 的第三方 SaaS（如个人 GitHub Token）。
   </Tab>
   <Tab title="成员授权（OAuth）">
-    成员通过 **OAuth 2.1** 流程各自授权；首次调用该服务器的工具时自动弹出授权窗口，授权服务器的发现（discovery）与动态客户端注册（DCR）按需进行，令牌按成员存储并在临近过期时自动刷新。适合支持 OAuth 的服务。
+    成员通过 **OAuth 2.1** 流程各自授权；首次调用该服务器的工具时自动弹出授权窗口，授权服务器的发现（discovery）与客户端注册按需进行：授权服务器提供注册端点时走**动态客户端注册（DCR）**；没有注册端点、但声明支持 Client ID Metadata Document 时，fc-safari 不注册客户端，而是把自己公开的 `/safari/oauth/client-metadata.json` 作为 `client_id`，由授权服务器按这个 URL 来取文档，你这一侧不需要任何额外配置。令牌按成员存储并在临近过期时自动刷新。适合支持 OAuth 的服务。
   </Tab>
 </Tabs>
 
@@ -180,7 +180,7 @@ MCP 服务器支持三种**授权模式**，决定凭证如何提供给服务器
 
 **凭证对话框里的授权操作**：对话框会根据其授权模式与您当前的凭证状态给出操作：
 
-- **成员授权（OAuth）**：先选择**执行环境**，再点击 **去授权**（未连接时）/ **重新授权**（已连接或已过期时）。可选择云端 Sandbox 或在线的 BYOC Runner，不能选择「自动」；OAuth 的发现、动态客户端注册、令牌交换和后续刷新都从所选环境发起。每次打开时会优先预选上次成功授权且仍在线的环境，否则预选绑定且在线的 Runner，最后回退到云端 Sandbox。OAuth 服务只能从内网访问时，请选择能访问它的 BYOC Runner。浏览器随后会弹出授权窗口（经 `/safari/credentials/oauth/initiate` 发起，返回授权链接后在弹窗中打开）。
+- **成员授权（OAuth）**：先选择**执行环境**，再点击 **去授权**（未连接时）/ **重新授权**（已连接或已过期时）。可选择云端 Sandbox 或在线的 BYOC Runner，不能选择「自动」；OAuth 的发现、动态客户端注册、令牌交换和后续刷新都从所选环境发起（授权服务器不支持 DCR 时改用 Client ID Metadata Document，见上文「认证」）。每次打开时会优先预选上次成功授权且仍在线的环境，否则预选绑定且在线的 Runner，最后回退到云端 Sandbox。OAuth 服务只能从内网访问时，请选择能访问它的 BYOC Runner。浏览器随后会弹出授权窗口（经 `/safari/credentials/oauth/initiate` 发起，返回授权链接后在弹窗中打开）。
 - **成员授权（密钥）**：点击 **填密钥**（未连接时）/ **更新密钥**（已连接时），弹出密钥录入弹窗，提交到 `/safari/credentials/secret`。
 - **撤销**：已有凭证时可点击 **撤销**（经 `/safari/credentials/revoke`）删除自己保存的凭证，状态回到「未连接」。
 
diff --git a/zh/ai-sre/memory.mdx b/zh/ai-sre/memory.mdx
index ca65324a1..3f40c4c68 100644
--- a/zh/ai-sre/memory.mdx
+++ b/zh/ai-sre/memory.mdx
@@ -74,7 +74,7 @@ AI SRE 每次新会话都从当前范围里已有的记忆文件读起。如果
 
 - **选范围**：顶部在 **个人** 和 **团队** 之间切换；选团队时再从下拉里挑一个您所属的团队。没有共享（账户级）记忆。范围和团队记在页面地址里，刷新或分享链接会打开同一组记忆。
 - **找记忆**：左侧列表按 偏好 / 做法 / 事实 / 教训 分组，组内按更新时间倒序；搜索框按名称、描述和文件名过滤。
-- **新建或编辑**：点 **新建记忆**，或在右侧点 **编辑**，填写名称、类型、一句话描述（什么时候用得上）和内容后保存。
+- **编辑**：在右侧点 **编辑**，修改名称、一句话描述（什么时候用得上）和内容后保存。这个页面不提供新建入口——记忆只能由对话产生（见[记忆如何产生](#记忆如何产生)）；**类型** 也不能在这里改，它在记忆被写下时就已定下，页面只是按类型分组展示、保存时原样写回。
 - **删除**：删除后，之后的对话不会再参考它。在对话里说忘掉一件事，效果相同。
 
 如果这条记忆在您编辑期间被别人或 AI 改过，保存时会提示，再保存一次会用您的版本覆盖；取消可以查看最新内容。
diff --git a/zh/changelog/changelog.mdx b/zh/changelog/changelog.mdx
index e202d22f0..dc5fe4eec 100644
--- a/zh/changelog/changelog.mdx
+++ b/zh/changelog/changelog.mdx
@@ -4,6 +4,29 @@ description: "本页面记录 Flashduty 产品的重要更新和功能发布"
 keywords: ["更新日志", "产品发布", "功能更新", "Flashduty", "版本记录"]
 ---
 
+<Update label="2026-10-09" description="📊 仪表盘内置时间变量与显示时区、RUM 查看器聚合分析、MCP Server 与 Go SDK 支持 OAuth 授权">
+
+### 仪表盘：内置时间变量与显示时区
+
+面板查询的表达式里现在可以直接写 `$__interval`、`$__rate_interval`、`$__range`、`$__auto` 四个**内置时间变量**，由面板自己的执行窗口与步长渲染成秒数——写 `$__range` 就不必再为「窗口有多长」手工维护一个变量。可用范围由数据源与查询方式决定：Prometheus 区间支持前三个，Loki 区间支持 `$__auto`、`$__interval`、`$__range`，Loki 即时支持 `$__auto` 与 `$__range`，其它组合在保存时即被拒绝。变量名以 `__` 开头是预留给内置变量的前缀，自定义变量不能再使用这个前缀。
+
+时间选择器弹层底部新增 **显示时区** 一行：可在 **默认**（跟随个人生效时区）、**浏览器时间**、**协调世界时** 以及按大洲分组的城市列表之间切换。它只作用于仪表盘，决定时间选择器的显示与解析、时序图的横轴与提示框时间，选择记在本浏览器。详见[仪表盘](/zh/monitors/dashboards/dashboards)。
+
+### RUM 查看器：聚合分析视图
+
+查看器顶部新增 **列表 / 分析** 两个页签。**分析** 视图把当前事件类型的事件按统计方式（计数、去重计数、平均值、求和、最小值、最大值与 P50 / P75 / P90 / P95 / P99）和分组字段汇总，提供时序图、排行、饼图、表格、分布图五种图表，支持在图上点击下钻回列表、拖选时间段放大，并可下载 PNG 或导出 CSV；分析视图的状态完整记录在 URL 参数里，复制链接即可复现同一次分析。详见 [RUM 查看器聚合分析](/zh/rum/explorer/analytics)。
+
+### MCP Server 与 Go SDK 支持 OAuth 授权
+
+远程 MCP 服务 `https://mcp.flashcat.cloud/mcp` 现在同时接受 APP Key 与 **MCP OAuth 访问令牌**：支持 MCP OAuth 的客户端（claude.ai、Claude Desktop、ChatGPT connectors 等）新建连接器时无需填写任何凭据，客户端会自动发现 Flashduty 授权服务器并在浏览器打开授权页，权限与 APP Key 相同；未携带凭据的请求会返回 `401`，并在 `WWW-Authenticate` 响应头中给出受保护资源元数据地址。Go SDK 相应提供 `NewClientWithAccessToken`，用 `Authorization: Bearer` 令牌鉴权，且不再注入 `app_key` 查询参数。详见 [MCP Server](/zh/developer/mcp-server) 与 [Go SDK](/zh/developer/go-sdk)。
+
+### CLI：自动化规则支持 `--timezone`
+
+`flashduty automation create` 与 `flashduty automation update` 新增 `--timezone`（IANA 时区名）：省略时 `create` 由服务端依次取调用者的成员时区、账户时区，最后回退 `Asia/Shanghai`；`update` 省略时保留已存储的时区，显式传空值会按 UTC 存储，并且会立即重算下次触发时间。同时 `alert list` 新增 `--alert-ids`、`--alert-keys`、`--asc`、`--by-updated-at`。详见[命令行工具](/zh/developer/cli)。
+
+</Update>
+
+
 <Update label="2026-09-30" description="📊 仪表盘同环比与多图联动、RUM 应用可关联代码仓库、私有化部署 License 到期提醒">
 
 ### 监控仪表盘：同环比对比、多图光标联动与点值筛选
diff --git a/zh/developer/cli.mdx b/zh/developer/cli.mdx
index 0eab27c31..ff845af10 100644
--- a/zh/developer/cli.mdx
+++ b/zh/developer/cli.mdx
@@ -42,6 +42,9 @@ irm https://static.flashcat.cloud/flashduty-cli/install.ps1 | iex
 | `FLASHDUTY_INSTALL_DIR` | 安装目录 | `/usr/local/bin`（Shell）、`~\.flashduty\bin`（PowerShell） |
 | `MIRROR_URL` | 覆盖安装脚本的 Release 资产下载镜像前缀，需以 `https://` 开头 | `https://static.flashcat.cloud/flashduty-cli` |
 | `FLASHDUTY_UPDATE_BASE_URL` | 覆盖 `flashduty update` 及自动更新检查的下载基础 URL | `https://static.flashcat.cloud/flashduty-cli` |
+| `INSTALLED_NAME` | 安装后的命令名，例如 `fduty` | `flashduty` |
+
+以自定义命令名安装的副本，在帮助、错误信息与 Shell 自动补全中都使用该名称；`flashduty update` 按正在运行的可执行文件名与所在目录就地替换自身，而不是另装一份默认命名的副本。
 
 ## 认证
 
@@ -61,6 +64,17 @@ CLI 按以下优先级查找 APP Key（高优先级在前）：
 2. `FLASHDUTY_APP_KEY` 环境变量
 3. 配置文件 `~/.flashduty/config.yaml`（由 `flashduty login` 写入）
 
+### 环境变量
+
+CLI 运行时读取以下环境变量（安装脚本识别的变量见上文「安装选项」）：
+
+| 环境变量 | 说明 | 默认值 |
+|----------|------|--------|
+| `FLASHDUTY_APP_KEY` | APP Key，覆盖配置文件中的 `app_key` | - |
+| `FLASHDUTY_BASE_URL` | API 地址，覆盖配置文件中的 `base_url` | `https://api.flashcat.cloud` |
+| `FLASHDUTY_UPDATE_BASE_URL` | 覆盖 `flashduty update` 及自动更新检查的下载基础 URL | `https://static.flashcat.cloud/flashduty-cli` |
+| `FLASHDUTY_NO_UPDATE_CHECK` | 设为 `1` 时关闭每日后台更新检查；CI 环境（`CI`、`GITHUB_ACTIONS`、`JENKINS_URL`、`GITLAB_CI`）下 CLI 也会跳过该检查 | - |
+
 ### 配置文件
 
 存储在 `~/.flashduty/config.yaml`，权限为 `0600`：
@@ -104,6 +118,7 @@ flashduty incident merge <target-id> --source <ids> # 将多个故障合并到
 flashduty incident timeline <id>        # 查看故障时间线
 flashduty incident alerts <id>          # 查看故障关联的告警
 flashduty incident similar <id>         # 查找相似的历史故障
+flashduty incident feed <id> [flags]   # 查看故障的分页动态流（时间线条目）
 ```
 
 `incident list` 常用过滤参数：
@@ -134,6 +149,8 @@ flashduty incident similar <id>         # 查找相似的历史故障
 
 时间格式示例：`5m`、`1h`、`24h`、`168h`、`2026-04-01`、`2026-04-01 10:00:00`、`1712000000`。
 
+`incident feed <id>` 按页读取故障的动态流（时间线条目）：`--asc` 让最旧的条目在前（默认最新在前），`--types` 按逗号分隔的条目类型过滤（如 `i_comm,i_assign`，条目的 `type` 字段给出实际类型），另有 `--limit`（最大 100）与 `--page`。
+
 #### 创建、更新与合并
 
 `incident create` 的 `--title`（3–200 字符）与 `--severity` 必填，在终端中缺少时会交互提示。指派方式有三种，可组合使用：`--assign`（成员 ID 列表）、`--assign-emails`（成员邮箱，逗号分隔；不匹配任何成员的邮箱会被忽略）、`--escalate-rule-id` 加 `--escalate-layer`（从该分派策略的指定环节（从 0 开始）起，指派该环节的人员；`--escalate-layer` 必须与 `--escalate-rule-id` 一起使用）。
@@ -202,6 +219,8 @@ flashduty incident comment-type-delete <comment-type-id>      # 删除
 flashduty incident comment-type-reorder <id> [<id2>...]       # 按给定顺序重排（需给出账户内全部类型 ID）
 ```
 
+写评论时用 `--comment-type-id` 附上账户级评论类型：`flashduty incident comment <id> [<id2>...] --comment-file <path> --comment-type-id <comment-type-id>`；评论类型 ID 来自 `flashduty incident comment-type-list`，省略该参数则写入不带类型的评论。
+
 ### change — 变更记录
 
 ```bash
@@ -668,6 +687,8 @@ flashduty alert-event list [flags]  # 列出告警事件（默认最近 1 小时
 
 两个 `list` 命令常用过滤参数：`--severity`（`Critical,Warning,Info`）、`--channel`（逗号分隔协作空间 ID）、`--integration`（逗号分隔的集成 ID）、`--since`/`--until`、`--limit`（最大 100）、`--page`。`alert-event list` 另有 `--integration-type`，按逗号分隔的**插件键**过滤（如 `AliCloud,Prometheus`）——注意它取插件键而非集成 ID，按集成 ID 过滤请使用 `--integration`。
 
+`alert list` 另有 `--alert-ids`（逗号分隔的告警 ID）、`--alert-keys`（逗号分隔的告警去重键）、`--asc`（按开始时间升序，默认最新在前）、`--by-updated-at`（把 `--since`/`--until` 应用到最近更新时间，而不是开始时间）；`alert-event list` 也支持 `--asc`（升序，默认最新在前），但没有 `--by-updated-at`。
+
 ### automation — AI SRE 自动化规则
 
 `automation` 命令组管理 AI SRE 的自动化规则——创建、查询、更新、删除、运行历史与触发，页面操作见[自动化](/zh/ai-sre/automations)。
@@ -693,18 +714,19 @@ flashduty automation fire <trigger_id>            # 经 HTTP POST 触发器触
 | `--at` | `HH:MM` 本地时间（按规则时区）；`hourly` 只取分钟（默认 0 分），`daily`/`weekly` 取小时与分钟（默认 09:00） |
 | `--weekday` | `weekly` 的星期：`sun`–`sat` 或 `0`–`7`（`0` 与 `7` 均表示周日，默认周一） |
 | `--cron-expr` | 精确的 5 段 cron 表达式（按规则时区解释），优先级高于 `--schedule` 辅助参数 |
+| `--timezone` | IANA 时区名（如 `Asia/Shanghai`），`--at` 与 `--cron-expr` 按该时区解释；省略时由服务端依次取调用者的成员时区、账户时区，最后回退 `Asia/Shanghai`。服务端无法加载的时区会被拒绝 |
 | `--disabled` | 以停用状态创建（默认启用） |
 | `--schedule-enabled` | 是否启用周期触发器（默认 `true`） |
 | `--http-post-trigger` | 创建并启用 HTTP POST 触发器；只传它且不带任何周期参数时，CLI 会写入占位 cron 并自动停用周期触发器 |
 | `--prompt` / `--prompt-file` | 任务提示词，二选一必填；`--prompt-file` 从文件读取（`-` 表示 stdin） |
 | `--environment-kind` / `--environment-id` | 运行环境：`cloud` 或 `byoc`（`byoc` 需配合 `--environment-id`）；留空表示自动 |
 
-`update` 可改字段：`--name`、`--prompt`/`--prompt-file`、`--schedule`/`--at`/`--weekday`/`--cron-expr`、`--enable`/`--disable`、`--enable-schedule`/`--disable-schedule`、`--enable-http-post-trigger`/`--disable-http-post-trigger`、`--rotate-http-post-token`、`--environment-kind`/`--environment-id`。个人 / 团队作用域创建后不可变更，`update` 不提供改作用域的参数。
+`update` 可改字段：`--name`、`--prompt`/`--prompt-file`、`--schedule`/`--at`/`--weekday`/`--cron-expr`/`--timezone`、`--enable`/`--disable`、`--enable-schedule`/`--disable-schedule`、`--enable-http-post-trigger`/`--disable-http-post-trigger`、`--rotate-http-post-token`、`--environment-kind`/`--environment-id`。个人 / 团队作用域创建后不可变更，`update` 不提供改作用域的参数。
 
 `runs` 支持 `--status`（`queued`/`running`/`retrying`/`succeeded`/`partial`/`failed`/`skipped`/`abandoned`）、`--trigger-kind`（`schedule`/`debug`/`http_post`）、`--since`/`--until`、`--page`、`--limit`。`fire` 用触发器的 Token 鉴权（`--token`，或环境变量 `FLASHDUTY_AUTOMATION_TRIGGER_TOKEN`），`--text` 传本次运行的上下文，`--data` 传完整 JSON 请求体（内联 JSON 或 `-` 读 stdin）。
 
 <Warning>
-**时区语义**：`--at` 与 `--cron-expr` 均按**规则时区**的本地挂钟时间理解——规则时区在创建时默认为调用者的成员时区，成员未设置时回退到账户时区。请直接传用户的本地时间，**不要**预先换算成 UTC。CLI 的 `create` / `update` 都没有 `--timezone` 参数：创建时如需固定其它时区，请改用生成命令 `flashduty safari automation-rule-create --timezone`；已创建规则的时区在 `update` 中不可更改。
+**时区语义**：`--at` 与 `--cron-expr` 均按**规则时区**的本地挂钟时间理解。`create` 省略 `--timezone` 时由服务端决定规则时区——依次取调用者的成员时区、账户时区，最后回退 `Asia/Shanghai`；需要固定其它时区时传入 IANA 名称（如 `--timezone Asia/Shanghai`），服务端无法加载的时区会被拒绝。`update` 省略 `--timezone` 时保留已存储的时区（包括历史遗留的空字符串），显式传空值 `--timezone ""` 会按 UTC 发送并存储，而不是账户默认时区。请直接传用户的本地时间，**不要**预先换算成 UTC。`update` 会立即重算下次触发时间，请以响应中的 `schedule_next_fire_at_ms` 为准。
 </Warning>
 
 ### insight — 洞察查询
diff --git a/zh/developer/go-sdk.mdx b/zh/developer/go-sdk.mdx
index 86f964446..8fc21b5ee 100644
--- a/zh/developer/go-sdk.mdx
+++ b/zh/developer/go-sdk.mdx
@@ -95,7 +95,7 @@ func main() {
 | `error` | `error` | 失败时为 `*ErrorResponse`,429 时为 `*RateLimitError` |
 
 <Tip>
-`app_key` 用于鉴权,SDK 会自动将其作为查询参数注入每个请求。请在 Flashduty 控制台的"推送集成"或团队配置中获取对应的 `app_key`。
+`app_key` 用于鉴权,SDK 会自动将其作为查询参数注入每个请求。请在 Flashduty 控制台的"推送集成"或团队配置中获取对应的 `app_key`。也可以改用 OAuth 访问令牌鉴权：用 `NewClientWithAccessToken` 创建客户端，见[创建客户端](#创建客户端)。
 </Tip>
 
 ## 创建客户端
@@ -132,6 +132,18 @@ client, err := flashduty.NewClient("YOUR_APP_KEY",
 **私有化部署**：用 `WithBaseURL` 将客户端指向您自有的 Flashduty 网关地址即可,其余调用方式完全不变。
 </Info>
 
+### OAuth 访问令牌
+
+除 `app_key` 之外，SDK 也支持用 OAuth 访问令牌鉴权：`NewClientWithAccessToken(accessToken, opts...)` 返回的客户端在 JSON、GET 与文件上传请求上都发送 `Authorization: Bearer <token>`，并且不再注入 `app_key` 查询参数——两者是二选一，不是叠加。令牌为空会直接返回错误（`flashduty: access token is required`）；除此之外它与 `NewClient` 完全一致，所有 `Option` 照旧可用，默认 Base URL、超时与 User-Agent 也相同。
+
+```go
+client, err := flashduty.NewClientWithAccessToken(accessToken,
+	flashduty.WithTimeout(10*time.Second),
+)
+```
+
+访问令牌由 Flashduty 的 OAuth 授权服务器签发，该服务器地址就是 API 基础地址（默认 `https://api.flashcat.cloud`）。客户端在 `/.well-known/oauth-authorization-server` 发现它，走授权码 + PKCE 流程换取令牌；令牌以所属账户的身份调用公开 API，权限与同一账户的 `app_key` 相同。令牌只在签发它的部署上有效，因此用访问令牌鉴权时不要用 `WithBaseURL` 指向其他网关。
+
 ## 服务与方法
 
 ---
diff --git a/zh/developer/mcp-server.mdx b/zh/developer/mcp-server.mdx
index f5581fa61..942cd6aae 100644
--- a/zh/developer/mcp-server.mdx
+++ b/zh/developer/mcp-server.mdx
@@ -38,9 +38,24 @@ Flashduty MCP Server 提供三种部署方式，您可以根据 MCP 客户端的
 
 <Tabs>
   <Tab title="远程服务（推荐）">
-    无需本地安装，直接连接官方远程实例 `https://mcp.flashcat.cloud/mcp`，使用 `Authorization: Bearer` 携带您的 APP Key 进行认证。这是最快的接入方式。
+    无需本地安装，直接连接官方远程实例 `https://mcp.flashcat.cloud/mcp`，这是最快的接入方式。远程服务支持两种认证方式，按客户端能力二选一：
 
-    以 Cursor 为例，在 MCP 配置中添加：
+    - **OAuth 登录**：客户端支持 MCP OAuth（如 claude.ai、Claude Desktop、ChatGPT connectors）时，新建一个自定义连接器，地址填 `https://mcp.flashcat.cloud/mcp`，不要填写任何凭据。客户端会自动发现 Flashduty 授权服务器并在浏览器打开授权页，您完成授权后即以自己的身份调用 Flashduty 开放 API，权限与 APP Key 相同。
+    - **APP Key**：客户端不支持 OAuth 时，改用 `Authorization: Bearer` 请求头携带您的 APP Key。
+
+    OAuth 登录只需要一个地址：
+
+    ```json
+    {
+      "mcpServers": {
+        "flashduty": {
+          "url": "https://mcp.flashcat.cloud/mcp"
+        }
+      }
+    }
+    ```
+
+    使用 APP Key 时，以 Cursor 为例，在 MCP 配置中添加：
 
     ```json
     {
@@ -121,7 +136,8 @@ Flashduty MCP Server 提供三种部署方式，您可以根据 MCP 客户端的
 | 项目 | 说明 |
 | --- | --- |
 | 认证（本地） | 通过 `FLASHDUTY_APP_KEY` 环境变量或 `--app-key` 命令行参数提供 Flashduty APP Key。 |
-| 认证（远程） | 通过请求头 `Authorization: Bearer <your_flashduty_app_key>` 提供 APP Key。 |
+| 认证（远程，OAuth） | 支持 MCP OAuth 的客户端无需提供凭据，在浏览器完成 Flashduty 授权即可。OAuth 只适用于远程服务，没有对应的环境变量或命令行参数。 |
+| 认证（远程，APP Key） | 通过请求头 `Authorization: Bearer <your_flashduty_app_key>` 提供 APP Key。 |
 | Streamable HTTP | `flashduty-mcp-server http`，端点为 `POST /mcp`。官方远程实例为 `https://mcp.flashcat.cloud/mcp`。 |
 | stdio | `flashduty-mcp-server stdio`，用于 Cursor、Claude Desktop 等本地客户端。 |
 
@@ -129,6 +145,20 @@ Flashduty MCP Server 提供三种部署方式，您可以根据 MCP 客户端的
   服务端不支持旧版独立 SSE 传输（单独的 `GET /sse` 端点），对 `GET` 请求会返回 `405 Method Not Allowed`。Streamable HTTP 已按 MCP 规范在 `POST` 上完成流式响应，请直接使用它；若客户端仅支持旧版 SSE，请升级客户端或在本地改用 stdio。
 </Warning>
 
+### OAuth 授权与发现
+
+远程服务同时接受 MCP OAuth 访问令牌与 APP Key：凭据以 `oauth:` 开头时按访问令牌处理，其余 Bearer 凭据一律按 APP Key 处理，两者权限相同。
+
+未携带任何凭据的请求会返回 `401`，并在 `WWW-Authenticate` 响应头中给出受保护资源元数据地址 `Bearer resource_metadata="<请求源>/.well-known/oauth-protected-resource<MCP 路径>"`。客户端据此发现元数据、进而发起 OAuth 授权流程。服务端提供以下元数据地址：
+
+| 元数据地址 | 说明 |
+| --- | --- |
+| `/.well-known/oauth-protected-resource` | 根路径，描述 `/mcp`。 |
+| `/.well-known/oauth-protected-resource/mcp` | 描述 `/mcp`。 |
+| `/.well-known/oauth-protected-resource/flashduty` | 描述兼容保留的 `/flashduty` 路径。 |
+
+元数据中的 `resource` 是请求源加上对应路径，`authorization_servers` 是 Flashduty API 地址，即本地通过 `FLASHDUTY_BASE_URL` 或 `--base-url` 配置的地址（默认 `https://api.flashcat.cloud`）——该地址同时充当授权服务器。访问令牌只在签发它的那个 API 地址上有效，因此远程 URL 上的 `base_url` 对 OAuth 令牌不生效，对 APP Key 仍然生效。
+
 ## 配置项
 
 ---
@@ -184,7 +214,7 @@ docker run -i --rm \
 
 ### 远程服务配置
 
-使用官方远程实例时，可在 URL 上追加查询参数动态配置工具集与只读模式：
+使用官方远程实例时，可在 URL 上追加查询参数动态配置工具集、只读模式与 Flashduty API 地址：
 
 ```json
 {
@@ -199,6 +229,12 @@ docker run -i --rm \
 }
 ```
 
+| 查询参数 | 说明 |
+| --- | --- |
+| `toolsets` | 以逗号分隔的启用工具集列表（如 `incidents,users,channels`）。 |
+| `read_only` | 设为 `true` 开启只读模式。 |
+| `base_url` | 覆盖 Flashduty API 地址，只对 APP Key 请求生效；OAuth 访问令牌只在签发它的 API 地址上有效，会忽略该参数。 |
+
 ### 输出格式（TOON）
 
 服务端支持 [TOON（Token-Oriented Object Notation）](https://github.com/toon-format/toon)格式输出工具结果。对于字段统一的对象数组（如成员列表、故障列表），TOON 相比 JSON 可显著减少 30%-50% 的 Token 消耗，且现代大模型可直接解析。设置 `FLASHDUTY_OUTPUT_FORMAT=toon`（或 `--output-format toon`）即可启用。
diff --git a/zh/monitors/dashboards/dashboards.mdx b/zh/monitors/dashboards/dashboards.mdx
index 0538f2d95..f5e627850 100644
--- a/zh/monitors/dashboards/dashboards.mdx
+++ b/zh/monitors/dashboards/dashboards.mdx
@@ -96,7 +96,7 @@ Prometheus 在工作台里同时跑区间与即时两条查询（Both），固
 
 | 配置项 | 说明 |
 |--------|------|
-| **查询表达式** | 支持 `${variable}` 模板；PromQL 数据源提示「输入 PromQL 查询」，其它数据源按类型提示 |
+| **查询表达式** | Prometheus / Loki 面板支持 `${variable}` 模板（大括号可省略，写成 `$variable` 也认），并可以使用 `$__interval` 等内置时间变量，见[内置时间变量](#内置时间变量)；MySQL 面板不支持变量插值；PromQL 数据源提示「输入 PromQL 查询」，其它数据源按类型提示 |
 | **多查询** | 查询用 `A`~`Z` 编号，只有时序图支持多查询；切到其它图表类型时必须明确选一个保留，其余会被删除 |
 | **序列名称** | 只有时序图显示，写每条序列展示的名字（图例、提示框、数据链接都用它），如 `{{instance}}` 会替换为该序列 `instance` 标签的值，留空显示全部标签；图例与提示框都隐藏时会提示该配置当前看不到效果 |
 | **最小步长** | 只有 Prometheus 的区间查询显示，是查询步长（step）的下限，默认 15s（对齐 Prometheus 默认抓取间隔）；切换到不接受该字段的数据源 / 查询方式时会被移除 |
@@ -121,6 +121,29 @@ Prometheus 在工作台里同时跑区间与即时两条查询（Both），固
 
 预览在面板编辑器里实时进行：改查询本身后按回车（或 `⌘/Ctrl + Enter`）才跑，改数据源、图表类型、时间这类上下文会自动重跑；显示层配置（单位、阈值、别名等）改了即时生效不用重查。**保存不要求预览成功**。新建仪表盘时如果还没选保存目录，预览区会提示「先选择保存位置才能预览」。
 
+### 内置时间变量
+
+面板查询的表达式里可以直接写内置时间变量，它们由这个面板自己的执行窗口与步长算出秒数——写 `$__range` 就不必为了「窗口有多长」再手工维护一个变量。
+
+| 变量 | 渲染成什么 |
+|------|-----------|
+| `$__interval` | 面板查询的步长（秒），如 `30s`；步长由时间范围与**最大点数**算出，并受**最小步长**约束 |
+| `$__rate_interval` | `步长 + 15s` 与 `60s` 里的较大者（15s 是假定的抓取间隔），适合直接喂给 `rate()` |
+| `$__range` | 面板执行窗口的长度（秒），如 24 小时的窗口渲染成 `86400s` |
+| `$__auto` | Loki 即时查询里是整个窗口的时长，按能整除的最大单位写成 `1h`、`5m`、`90s` 或 `90500ms`；Loki 区间查询里与 `$__interval` 一样给步长秒数 |
+
+写法与变量一致：`${__interval}` 与 `$__interval` 都接受（大括号可省略），但内置时间变量不吃格式化后缀，写成 `${__interval:raw}` 之类的会被判为非法。
+
+可用范围由数据源与查询方式决定，下表之外的组合（包括 MySQL）都用不了；这种组合在保存校验时就会被拒（保存弹窗按 JSON Pointer 列出原因），面板运行起来也不会取数：
+
+| 数据源 · 查询方式 | 可用的内置时间变量 |
+|------------------|------------------|
+| Prometheus · 区间 | `$__interval`、`$__rate_interval`、`$__range` |
+| Loki · 区间 | `$__auto`、`$__interval`、`$__range` |
+| Loki · 即时 | `$__auto`、`$__range` |
+
+内置时间变量只在**面板查询的表达式**里存在：模板变量的候选查询、面板查询的 args 都解析不了它们，保存时同样会被拒。
+
 ## 模板变量
 
 变量栏在画布上方，用全站统一的筛选控件呈现：每个变量是一个常驻筛选项，单选 / 多选由变量配置决定。编辑态在变量栏末尾多了 **管理变量**（变量管理抽屉），可以增、改、删全部变量。
@@ -133,9 +156,9 @@ Prometheus 在工作台里同时跑区间与即时两条查询（Both），固
 | **Custom** | 手写的固定候选列表 | 候选项（每行一个，`显示名=值` 或纯值） |
 | **Query** | 对数据源查询得到候选 | 数据源（固定数据源或数据源变量）+ 查询方式 + 候选刷新时机 |
 
-**Query 变量的候选查询**按数据源类型分三种：Prometheus 需要填 `label`（必填）、`metric`（选填），并可用「标签过滤」逐条添加 `= / != / =~ / !~` 条件；SQL 数据源填表达式与取值字段；Loki 填表达式与字段。抽屉里的 **预览候选** 会带上当前配置试拉一次，用来确认数据源与查询写对了，失败时在变量旁显示错误原因。
+**Query 变量的候选查询**按数据源类型分三种：Prometheus 需要填 `label`（必填）、`metric`（选填），并可用「标签过滤」逐条添加 `= / != / =~ / !~` 条件；SQL 数据源填表达式与取值字段；Loki 填表达式与字段——字段必须是流的标签名（`^[A-Za-z_][A-Za-z0-9_]*$`，点分字段名、JSON 里的字段名都不行），候选值取自表达式选中的那些流在**仪表盘时间范围内**的该标签取值，表达式留空即在所有流上取；表达式里可以用仪表盘变量，但不能用内置时间变量（它们只在面板查询的表达式里可用，见[内置时间变量](#内置时间变量)）。抽屉里的 **预览候选** 会带上当前配置试拉一次，用来确认数据源与查询写对了，失败时在变量旁显示错误原因。
 
-**名称与引用**：变量名必须匹配 `^[A-Za-z_][A-Za-z0-9_]{0,63}$`，与已有变量重名会被拒绝；改名是前端的原子重构——会把全部 `${旧名}` 引用（查询表达式、查询 args、数据源变量引用、变量自身的候选查询）一起改掉，改动前提示「将同步更新 N 处引用」。被引用着的变量不能删除，会提示「变量 x 被 N 处引用，请先清理引用」。一个仪表盘最多 20 个变量。
+**名称与引用**：变量名必须匹配 `^[A-Za-z_][A-Za-z0-9_]{0,63}$`，也不能以 `__` 开头（这个前缀留给内置时间变量），与已有变量重名会被拒绝；改名是前端的原子重构——会把全部 `${旧名}` 引用（查询表达式、查询 args、数据源变量引用、变量自身的候选查询）一起改掉，改动前提示「将同步更新 N 处引用」。被引用着的变量不能删除，会提示「变量 x 被 N 处引用，请先清理引用」。一个仪表盘最多 20 个变量。
 
 **选择模式与「全部」**：
 
@@ -151,6 +174,15 @@ URL 是变量选择的唯一事实源：单值写 `var-<名字>=值`，多值重
 
 - **时间范围**由 URL 的 `from` / `to` 决定，可以是绝对毫秒，也可以是 `now-*` 相对表达式；没有 URL 参数时用盘上保存的**默认时间范围**（新建时为「过去 1 小时」，保存时可在 15m / 30m / 1h / 3h / 6h / 12h / 24h / 7d 八档里固化，且必须以「现在」为终点）。
 - **自动刷新**：盘上保存的频率是 关闭 / 30s / 1m / 5m 四档；查看页的刷新控件提供更细的档位（关闭 / 10s / 30s / 1m / 5m / 15m / 30m / 1h），选择后会写进 URL 的 `refresh` 参数，并立即刷新一次，不必等满一个周期。相对时间档在刷新时会跟着「现在」滚动窗口。左侧的刷新按钮随时可以手动刷新，页面还有查询在跑时图标转圈。
+- **显示时区**：时间选择器弹层底部是一行时区设置——左侧是当前时区（城市 + 缩写）与 UTC 偏移，右侧 **修改时区** 按钮打开一个可搜索的城市列表（可按城市名、缩写或 IANA 名搜），列表按大洲分组，首组 **常用** 是下表的三个特殊项。它只作用于仪表盘（查看页、编辑态、面板编辑器与面板全屏），决定时间选择器的显示与解析、时序图的横轴刻度与提示框时间、时间范围文案；真正按时区渲染的是绝对时刻，相对档（如「过去 1 小时」）表示的是时长，不随时区平移。显示时区与个人生效时区不一致时，时间选择器上会带该时区的缩写提醒。
+
+| 常用项 | 含义 |
+|--------|------|
+| **默认** | 跟随个人设置的生效时区 |
+| **浏览器时间** | 跟随当前浏览器所在时区 |
+| **协调世界时** | 固定使用 UTC |
+
+选择记在本浏览器（localStorage 键 `monit.dashboards.timeZone`），下次打开沿用；选 **默认** 就清掉这条记录，回到个人设置的生效时区。
 
 ## 同环比对比
 
diff --git a/zh/monitors/explore/explore.mdx b/zh/monitors/explore/explore.mdx
index c34290902..04181238d 100644
--- a/zh/monitors/explore/explore.mdx
+++ b/zh/monitors/explore/explore.mdx
@@ -78,7 +78,32 @@ keywords: ["查询工作台", "Explore", "即时查询", "Prometheus", "Loki", "
 | 项目 / 日志库 | SLS | 必填，拼入请求地址 |
 | 地域 / 日志主题 / 语法 | 腾讯云 CLS | 必填 |
 | PowerSQL | SLS | 开启 PowerSQL 模式 |
-| 时间宏 | SQL 数据源 | `$__timeFilter(列名)`、`$__timeFrom()`、`$__timeTo()` 让时间范围生效 |
+| 时间宏 | SQL 数据源 | `$__timeFilter(列名)`、`$__timeFrom()`、`$__timeTo()`；只有 SQL 里写了时间宏，页面选择的时间范围才会作用到这条查询，没有宏时 SQL 原样执行 |
+| 内置时间变量 | Prometheus / Loki | `$__interval`、`$__rate_interval`、`$__range`、`$__auto` 让查询跟随页面选择的时间范围与 step，见下节 |
+
+### 内置时间变量
+
+Prometheus 和 Loki 查询支持内置时间变量：变量在查询执行前由本次查询的时间窗口与 step 渲染成具体时长，因此查询会跟随页面选择的时间范围，不必把时间写死。变量名以 `$` 起头，也可以写成 `${...}` 形式。
+
+| 变量 | 渲染成的值 |
+|------|-----------|
+| `$__interval` | 本次查询的 step（秒），例如 `30s`；与下发给 Edge 的 step 相同，取后端按数据点数反推的值与 **最小 step** 中较大者 |
+| `$__rate_interval` | step 加 15 秒、且不低于 60 秒，例如 `60s`；15 秒是 `$__rate_interval` 假定的抓取间隔 |
+| `$__range` | 本次查询的时间窗口长度（秒），最小 1 秒，例如 `3600s` |
+| `$__auto` | Loki 区间查询里等于 step（例如 `30s`）；Loki 即时查询里本应取整个时间窗口的最大可整除单位（例如 `1h`），但即时查询不带起始时间，用不了（见下） |
+
+可用范围由数据源和查询类型共同决定：
+
+| 查询 | 可用的内置时间变量 |
+|------|------------------|
+| Prometheus 区间（**仅图表**，以及 **图表 + 原始** 的图表部分） | `$__interval`、`$__rate_interval`、`$__range` |
+| Prometheus 即时（**仅原始**，以及 **图表 + 原始** 的原始部分） | 无 |
+| Loki 区间 | `$__auto`、`$__interval`、`$__range` |
+| Loki 即时 | 无 |
+
+- 即时查询里没有能用的内置时间变量：Prometheus 即时查询不接受区间变量，Loki 即时查询不带起始时间，`$__auto`、`$__range` 这类需要窗口长度的变量渲染不出来。写了这些变量，结果区会提示「查询或参数不合法，请检查后重试。」，并在提示下方给出后端原因（例如 `built-in variable "__rate_interval" is not available in prometheus instant queries`、`built-in variable "__range" requires from_ms`）。
+- Prometheus 默认的 **图表 + 原始** 会同时下发一条区间查询和一条即时查询，两者共用同一份查询语句：`$__rate_interval` 在图表那一半正常渲染，在原始那一半失败。要在这类查询里用 `$__rate_interval`，把查询类型切到 **仅图表**。
+- 内置时间变量与仪表盘的内置时间变量同名，由同一套渲染逻辑按本次查询的时间窗口和 step 计算。除上表中的变量名外，其他 `$` 文本（包括 `${名称}`、`$1` 这类正则捕获组引用）原样送给数据源，不做替换——工作台没有仪表盘变量。
 
 ## AI 查询助手
 
diff --git a/zh/openapi/go-sdk.mdx b/zh/openapi/go-sdk.mdx
index c52542abd..95f7e87f5 100644
--- a/zh/openapi/go-sdk.mdx
+++ b/zh/openapi/go-sdk.mdx
@@ -10,7 +10,7 @@ keywords: ["Go SDK", "go-flashduty", "Golang", "Open API", "客户端", "SDK"]
 如果你的服务使用 Go 编写，推荐用该 SDK 调用 Flashduty，而不是手写 HTTP 请求 —— 你可以直接拿到类型化的请求参数、响应结构和错误码，省去手动拼装 JSON 和解析响应的工作。
 
 <Tip>
-SDK 与 Open API 共用一套接口语义。本文档每个接口的请求/响应结构，都能在 SDK 中找到对应的类型。鉴权同样使用 [APP Key](/zh/openapi/introduction#认证方式)。
+SDK 与 Open API 共用一套接口语义。本文档每个接口的请求/响应结构，都能在 SDK 中找到对应的类型。鉴权同样使用 [APP Key](/zh/openapi/introduction#认证方式)，也可以改用 OAuth 访问令牌，见[配置选项](#配置选项)。
 </Tip>
 
 ## 安装
@@ -91,6 +91,16 @@ client, err := flashduty.NewClient("YOUR_APP_KEY",
 )
 ```
 
+### OAuth 访问令牌
+
+除 `NewClient` 外，SDK 还提供 `NewClientWithAccessToken(accessToken, opts...)`：它用 OAuth 访问令牌鉴权，在 JSON、GET 与文件上传请求上发送 `Authorization: Bearer <token>`，不再注入 `app_key` 查询参数；令牌为空直接返回错误，可选项与 `NewClient` 完全相同。令牌由 Flashduty 的 OAuth 授权服务器签发（地址即 API 基础地址），且只在签发它的部署上有效。
+
+```go
+client, err := flashduty.NewClientWithAccessToken(accessToken,
+	flashduty.WithTimeout(10*time.Second),
+)
+```
+
 ## 错误与限流
 
 SDK 返回类型化的错误，可用 `errors.As` 解析，也提供便捷断言函数（自动穿透被包装的错误）：
diff --git a/zh/rum/explorer/analytics.mdx b/zh/rum/explorer/analytics.mdx
new file mode 100644
index 000000000..c855e7009
--- /dev/null
+++ b/zh/rum/explorer/analytics.mdx
@@ -0,0 +1,139 @@
+---
+title: "RUM 查看器聚合分析"
+sidebarTitle: "聚合分析"
+description: "在 Flashduty RUM 查看器的分析视图中用统计方式、分组与多种图表汇总 RUM 事件，并从聚合结果下钻到原始事件。"
+keywords: ["RUM", "数据查看器", "聚合分析", "图表", "下钻"]
+---
+
+RUM 查看器提供**列表**与**分析**两种查看方式。**分析**视图把当前事件类型的事件按你选定的统计方式与分组汇总成图表，用于回答「哪个页面的加载最慢」「错误集中在哪些浏览器」这类问题；需要看单条事件时再切回列表。
+
+## 列表与分析
+
+查看器顶部有两个页签：**列表**逐条展示事件，**分析**把事件汇总成图表。两者共用同一套事件类型、查询条件和时间范围，切换页签不会丢失当前的筛选条件。
+
+分析视图记录在 URL 的 `mode` 参数中：`mode=analytics` 打开分析，省略 `mode` 则打开列表，因此已有的查看器链接仍然指向列表。复制分析视图的链接即可与他人分享同一张图表。
+
+| 页签 | `mode` 取值 | 展示内容 |
+|------|-------------|----------|
+| **列表** | 省略 | 事件趋势图与事件表格，逐条下钻事件详情 |
+| **分析** | `analytics` | 按统计方式与分组聚合后的图表 |
+
+## 选择统计方式与字段
+
+分析视图顶部的**统计**一行决定算什么：先选**统计方式**，再按需选择**统计字段**，然后选择**分组字段**。
+
+### 统计方式
+
+| 统计方式 | `agg` 取值 | 说明 |
+|----------|-----------|------|
+| **计数** | `count` | 事件条数，不需要字段 |
+| **去重计数** | `count_distinct` | 字段去重后的取值个数 |
+| **平均值** | `avg` | 数值字段的平均值 |
+| **求和** | `sum` | 数值字段的合计 |
+| **最小值** | `min` | 数值字段的最小值 |
+| **最大值** | `max` | 数值字段的最大值 |
+| **P50 / P75 / P90 / P95 / P99** | `p50` / `p75` / `p90` / `p95` / `p99` | 数值字段的近似分位数，用于观察长尾 |
+
+除**计数**外，所有统计方式都需要先选择**统计字段**。选择平均值、求和、最小值、最大值或任一分位数时，字段选择器只列出数值类型字段；时间戳字段（单位 `unix_millisecond`）不参与数值统计，因此不在候选之列。选择**去重计数**时，字段选择器列出可作为分组的标量 facet 字段。切换事件类型后，如果原字段在新的事件类型上不存在，选择器会提示重新选择字段，而不会把问题悄悄变成计数。
+
+图例、表格列与 CSV 表头使用同一套命名：**计数**与**分布图**的统计值显示为「事件数」，其他统计方式显示为「统计方式（字段名）」，例如「平均值（view_loading_time）」。
+
+### 分组字段
+
+**分组字段**决定按什么拆开。选定后，图表为每个分组取值单独展示，并在旁边出现**取前**选择器，可选 `5`、`10`、`20`、`50` 组，默认 `10` 组。分组字段可以清空（占位文案为「不分组」），清空后结果汇总为单个数值。分组字段只列出标量 facet 字段；数组字段一条事件持有多个取值，无法作为分组依据。
+
+**分布图**不使用分组字段，该选择器不显示。
+
+## 图表类型
+
+分析视图提供五种图表类型：
+
+| 图表类型 | `viz` 取值 | 说明 |
+|----------|-----------|------|
+| **时序图** | `timeseries` | 按时间桶展示统计值随时间的走势，为默认图表类型 |
+| **排行** | `toplist` | 各分组的统计值从大到小排行 |
+| **饼图** | `pie` | 各分组的占比，需满足下方适用范围 |
+| **表格** | `table` | 分组与统计值两列明细 |
+| **分布图** | `distribution` | 一个数值字段按区间分桶后的事件数直方图 |
+
+**时序图**额外提供**画法**选择器，可在**折线**（`line`）与**柱状**（`bar`）之间切换，默认柱状。选择分组字段、画法为柱状、且统计方式为**计数**或**求和**时，柱状图按分组堆叠，柱高即为合计值；其他统计方式（平均值、分位数、去重计数）不能堆叠，因为各分组的值不可相加。
+
+未选择分组字段时，**排行**与**表格**直接展示该统计值本身，而不是画一根柱子；**时序图**仍按时间展示一条曲线。
+
+## 点击下钻
+
+点击图上的元素会弹出下钻菜单，用于从聚合结果回到明细事件：时序图的某个时间点、排行或表格的某一行、饼图的某个扇区、分布图的某个区间。
+
+| 菜单项 | 作用 |
+|--------|------|
+| **在列表中查看** | 切换到列表页签，只展示该元素背后的事件 |
+| **拆分「值」** | 只保留点击的分组取值，并打开分组字段选择器，以便继续按下一层字段拆分；分布图的区间没有分组，此时该项为**筛选为「区间」**，只保留落在该区间的事件 |
+| **排除「值」** | 从结果中排除点击的分组取值 |
+
+下钻写入的是与左侧 facet 列表相同的筛选条件，因此搜索框中会显示这些条件，也可以按同样的方式撤销。时序图的时间点下钻会把列表的时间范围收敛到该时间桶，分布图的区间下钻会把列表按该数值区间筛选。
+
+从**在列表中查看**进入列表后，列表顶部出现**返回分析**条，提示「列表已按下钻的条件筛选，返回时恢复原来的条件」。点击即可回到分析视图，并恢复下钻前的查询条件。手动切换页签会结束这次下钻，返回条随之消失。
+
+<Note>
+图上显示为「(空值)」的空值分组暂不支持下钻筛选，对它的**拆分**与**排除**均不可用。
+</Note>
+
+## 缩放时序图
+
+在**时序图**上拖选一段时间即可放大到该时间窗。缩放只作用于这张图，查看器自身的时间范围保持不变。放大后图表右上角显示**已放大**徽标与当前窗口范围，点击其中的关闭按钮（**重置缩放**）恢复全量时间范围。切换事件类型、修改查询条件或调整查看器时间范围都会重置缩放。
+
+## 下载图片与导出 CSV
+
+图表右上角提供两个导出按钮：
+
+| 按钮 | 说明 |
+|------|------|
+| **下载图片** | 将当前图表导出为 PNG，以 2 倍像素密度渲染。表格类型与未选择分组字段的单个数值没有图形可导出，此时按钮不可用 |
+| **CSV** | 将当前图表的数据导出为 CSV，文件名形如 `rum-<事件类型>-analytics-<时间戳>.csv`。时序图按「时间、分组、统计值」逐行导出（长表格式）；其他图表每个分组一行 |
+
+结果为空时两个按钮均不可用。
+
+## 分布图
+
+**分布图**按一个数值字段的取值区间分桶，展示每个区间内的事件数，不使用统计方式与分组字段。
+
+- 分桶范围从字段的最小值一直覆盖到 `P99`：RUM 的耗时类数据长尾很长（几千次 1 秒的加载中夹着一次 46 秒），若一直分到最大值，绝大多数取值会被挤进第一根柱子。
+- 分桶宽度自动取 `1`、`2` 或 `5` 乘以 10 的幂，使区间边界落在整齐的数值上，整体约切分 40 个桶。
+- 高于 `P99` 的取值全部并入最后一个开放区间，图上与 CSV 中显示为「≥ 边界值」。
+- 区间按字段自身的单位显示，例如耗时字段的 `1s–1.5s`。
+
+## 饼图的适用范围
+
+饼图必须同时满足两个条件，否则图表区域提示「饼图需要分组，且只适用于计数和求和」：
+
+- 选择了**分组字段**；
+- 统计方式为**计数**或**求和**——只有这两种统计方式的分组值可以相加为整体，去重计数、平均值与分位数都不满足这一前提。
+
+## 采样估算
+
+当数据量较大、结果由抽样数据估算得到时，图表右上角显示**采样估算**徽标，悬停提示「数据量较大，结果按抽样数据估算」。
+
+## URL 参数
+
+分析视图的状态保存在 URL 中，复制链接即可复现同一次分析。参数取值非法时会回退到默认值。
+
+| 参数 | 含义 | 取值 | 默认值 |
+|------|------|------|--------|
+| `mode` | 查看方式 | `analytics` | 省略即列表 |
+| `agg` | 统计方式 | `count`、`count_distinct`、`avg`、`sum`、`min`、`max`、`p50`、`p75`、`p90`、`p95`、`p99` | `count` |
+| `aggField` | 统计字段 | 字段名 | 无 |
+| `groupBy` | 分组字段 | 字段名 | 无（不分组） |
+| `top` | 取前 N 组 | `5`、`10`、`20`、`50` | `10` |
+| `viz` | 图表类型 | `timeseries`、`toplist`、`pie`、`table`、`distribution` | `timeseries` |
+| `style` | 时序图画法 | `line`、`bar` | `bar` |
+
+## 下一步
+
+<CardGroup cols={2}>
+  <Card title="RUM 查看器概览" icon="compass" href="./overview">
+    了解查看器的核心功能
+  </Card>
+  <Card title="数据查询指南" icon="magnifying-glass" href="./data-query">
+    掌握查询语法，精准定位数据
+  </Card>
+</CardGroup>
diff --git a/zh/rum/explorer/overview.mdx b/zh/rum/explorer/overview.mdx
index 8fd26c141..0e8677c61 100644
--- a/zh/rum/explorer/overview.mdx
+++ b/zh/rum/explorer/overview.mdx
@@ -22,6 +22,9 @@ Flashduty RUM **查看器**（RUM Explorer）是一款强大的数据分析工
   <Card title="数据搜索过滤" icon="filter">
     通过搜索栏和可视化类型选择，对 RUM 事件进行精确过滤和筛选
   </Card>
+  <Card title="聚合分析" icon="chart-line" href="./analytics">
+    按统计方式与分组把 RUM 事件汇总成图表，并从聚合结果下钻到原始事件
+  </Card>
 </CardGroup>
 
 ## 价值与优势
@@ -149,6 +152,9 @@ EventTable 的列选择器也已经支持以下小程序专属字段作为可查
   <Card title="数据查询指南" icon="magnifying-glass" href="./data-query">
     掌握查询语法，精准定位数据
   </Card>
+  <Card title="聚合分析" icon="chart-line" href="./analytics">
+    用统计方式、分组与图表汇总 RUM 事件
+  </Card>
   <Card title="分布式追踪" icon="diagram-project" href="../best-practices/distributed-tracing">
     了解分布式追踪最佳实践
   </Card>
diff --git a/zh/rum/quickstart/app-management.mdx b/zh/rum/quickstart/app-management.mdx
index 19e495b55..5264f1fc4 100644
--- a/zh/rum/quickstart/app-management.mdx
+++ b/zh/rum/quickstart/app-management.mdx
@@ -83,6 +83,8 @@ RUM 应用是承载前端性能监控数据的容器，用于采集、存储和
 
 每个平台的 SDK 配置页面都会自动填入当前应用的 `applicationId` 和 `clientToken`，您可以直接复制代码到项目中使用。
 
+Web、Android、iOS、Flutter、HarmonyOS 与微信小程序的 SDK 配置页都提供 **允许远程配置** 开关，默认为开启。开启时生成的初始化代码会带上 `remoteConfigurationEnabled: true`（Android 与 HarmonyOS 端为对应的 `.setRemoteConfigurationEnabled(true)` 初始化参数）；关闭后该行不会出现在代码中，客户端也就不再请求远程配置。开关只影响生成的代码——配置本身还需在 [远程配置](#远程配置) 页签中启用并发布，发布前客户端仍使用代码中的配置。
+
 <Warning>
 在应用管理中修改 SDK 配置并不会实时生效到已集成的客户端。所有配置更改需要在您的前端代码中更新并重新部署才能生效。
 </Warning>
@@ -133,6 +135,7 @@ flashcatRum.init({
 | `service` | 服务名称，所有事件默认带上此标签 | 仅支持英文、数字、下划线；最长 24 字符 | 无 |
 | `version` | 版本号，便于数据分析时筛选 | 仅支持英文、数字、`.`；最长 24 字符 | `1.0.0` |
 | `sessionSampleRate` | 会话采样率（百分比） | 整数，范围 `0 ~ 100` | `10` |
+| `remoteConfigurationEnabled` | 是否在生成的初始化代码中启用远程配置 | 无 | 开启 |
 
 <Note>
 表单字段填写完毕后，无需点击保存——预览代码会随输入实时更新。`applicationId` 和 `clientToken` 由系统自动填入，无需手动配置。
@@ -161,7 +164,8 @@ flashcatRum.init({
   service: "<SERVICE_NAME>",
   env: "<ENV_NAME>",
   version: "1.0.0",
-  sessionSampleRate: 10
+  sessionSampleRate: 10,
+  remoteConfigurationEnabled: true // 应用控制台中发布的配置
 });
 ```
 </Step>
@@ -391,7 +395,7 @@ iOS、Android、Flutter 与微信小程序应用同样展示「远程配置」
 
 - 变更历史保留最近 50 个版本，更早的版本将自动清理且无法回滚。每条历史展示操作人与变更说明（如填写）。你可以将任一保留中的历史版本回滚为当前配置，回滚同样会生成新版本。
 - 回滚产生的新版本带内容等价标注：若其配置内容与历史上某个更早版本完全相同，历史列表会标注该版本与那个最早相同内容的版本等价——版本号跳变不代表配置内容发生了变化。
-- 发布面板按最近到达的采样会话展示**采用情况**与采用进度，状态标签为「等待新会话」（刚发布，尚无会话按当前版本开始）、「生效中」（部分会话已切换）、「最近会话已切换」（窗口内到达的会话均已在当前版本）、「未收到配置」（最近会话均未上报配置版本，常见为 SDK 未开启远程配置或版本过低）、「暂无数据」（窗口内暂无新会话）、「未检测到生效」（发布已超过约 8 小时仍未切到当前版本）；进度条中未上报配置版本的会话段标为「未上报」（SDK 版本过低或尚未拉取过配置）。点击**查看明细**可跳转到事件浏览器查看这批样本会话。窗口内采样会话少于 20 条时，面板只展示实际会话数、不估算占比（占比由采样率放大估算，样本过少时结果没有意义）。
+- 发布面板按最近到达的采样会话展示**采用情况**：方格图以 10×10 共 100 格表示各版本的会话占比，每格代表 1% 的会话（按各版本采样率估算），右侧明细表逐组列出**版本**、**占比**与**会话数**。上方数字区展示**活跃会话**、**已上报**、**未上报**与**当前版本采用率**四项统计，其中当前版本采用率在窗口内到达的会话全部切换后附带勾选的「最近会话已切换」标记。各组按以下规则归并：当前版本单独一组并标注「当前版本」，占比最高的两个旧版本各自一组，其余旧版本合并为「更早版本」，未上报配置版本的会话单独列为「未上报」（SDK 版本过低或尚未拉取过配置）。状态标签为「等待新会话」（刚发布，尚无会话按当前版本开始）、「生效中」（部分会话已切换）、「最近会话已切换」（窗口内到达的会话均已在当前版本）、「未收到配置」（最近会话均未上报配置版本，常见为 SDK 未开启远程配置或版本过低）、「暂无数据」（窗口内暂无新会话）、「未检测到生效」（发布已超过约 8 小时仍未切到当前版本）。点击**查看会话明细**可跳转到事件浏览器查看这批样本会话；在版本卡片上点击**查看配置**可展开当前生效版本的完整内容，再次点击**收起配置**收起。窗口内采样会话少于 20 条时，面板不画方格图、只展示各组的实际会话数、不估算占比（占比由采样率放大估算，样本过少时结果没有意义）。
 
 ## 删除应用