Repository navigation
Conversation
…and refresh the Kubernetes and Hosts images
Add Kubernetes Map, host Metrics, and Refresh alert group Audit screenshots.
A Kubernetes object's Where shows its cluster, and its namespace when it has one, so a namespace shows its cluster. A new section says when each source is checked, what Last seen means, that the page doesn't refresh by itself, and when a resource that stops sending data leaves the list.
…d list the 17 tools it now exposes OAuth sign-in is the main way to connect: a client needs only the URL, and the person signs in, picks one workspace on the consent page, and clicks Allow. Tabs for Claude Code, Cursor, VS Code, and Claude (web and Desktop) follow each client's own OAuth flow, so the mcp-remote workaround is gone. A Personal API Key with x-workspace-id stays as the option for scripts and for clients that can't do OAuth. The tool list matches catalogTools in @kloudmate/kloudmate-ai-agent 2.6.0. Verify the connection and Security and scope cover token renewal, lost workspace access, and disconnecting from the client. Matches backend develop (223e1bb23, ai-agent 2.6.0) and frontend develop (b5313509). Backend master has neither yet, so don't publish ahead of prod.
….6.0 MCP tools One search_traces step grouped by serviceName (window 7d, view detail) replaces the two per-service steps, and check_incidents and check_slos replace search_incidents and search_slos. The tools compute error rates and give latency in milliseconds, so the prompt drops its arithmetic and unit rules. It also skips disabled SLOs, because check_slos lists them. The YAML passes fromPortable and validatePublish, leaving only the Slack channel for the reader to pick. The three calls ran against the local MCP server; search_traces took 11.4s over 7 days, inside the 30s MCP Tool Call timeout.
… with its Tools dialog The tab lists the connections that can be used for AI. Any role can connect a tool server there, and a new one starts Private. Only a Developer who created a server can share it. The Tools dialog covers Enabled, the three approval modes (Default runs Read tools and asks before any other), and the server-set tool labels. New screenshots replace the removed Connect custom server layout. In a workflow's Ask KloudMate Assistant step, the Assistant uses only the enabled tools labeled Read and runs them without asking, so actions.mdx no longer says Ask before run tools are off. It also links to the renamed heading.
…n the workspace can use a Workspace one Both connect paths, the credential form and OAuth sign-in, create a connection used only for AI as Private and every other connection as Workspace. A Workspace connection can be used by anyone in the workspace and managed by the Developer role. Only a Developer who created a private connection can share it.
…s have five sign-in options, and members manage their own AI connections AWS connections authenticate with access keys only, so the Connections pages, the workflow docs, and the two AWS guides now point to the IAM user's policy instead of a role. Custom MCP Server lists None, Bearer token, Basic auth, API key, and OAuth, and can be used for Workflows and AI. An OAuth server is saved as Pending, then opens the server's own sign-in, and needs a reconnect after its URL changes. Any connection used only for AI can be private. Atlassian takes a personal API token with the account email, or a service account API key, and needs API tokens allowed in Rovo. Anyone can edit, re-check, reconnect, and delete a private AI connection they created; sharing it or adding a capability needs the Developer role. The API keys page gives the MCP server two ways in: a Personal API Key, or OAuth sign-in. Matches backend develop (cc9b379a6, f86be246e, 0e44b1b60), which isn't on master yet, so don't deploy ahead of prod.
…etes The install scripts always fetch the latest release, so rerunning the install command upgrades the agent. Linux keeps agent.yaml; the Windows installer rewrites it with only the API key and endpoint. Kubernetes uses helm upgrade --install, and each chart version rolls the matching agent image. Co-Authored-By: Ansh Garhewal <anshdevsin@gmail.com>
docs(agent): add upgrade steps for Linux, Windows, Docker, and Kubernetes
…cident, opened by hand, by a routing rule, or by an escalation step Three pages under Incident Management: what responders see in a room and what each card button does, how to point war rooms at a Slack or Microsoft Teams channel and attach a conference call from Microsoft 365, Google Meet, or Zoom, and what to do when a room will not open. Slack needs the Workflows capability and a reconnect before it can create a channel and invite people, which is where the Reconnect banner comes from. Two behaviors are called out where the reader meets them: reopening an incident does not cancel an archive that is already scheduled, and a Teams group chat is closed by its summary rather than archived.
…nnection granted for Meetings Corrects the setup and troubleshooting pages against develop. Zoom connects by installing the KloudMate app, creates every meeting as the installing account, and revokes the connection when the app is removed. Microsoft 365 is granted for Notifications and Meetings in one go, and the meeting picker lists whatever holds the Meetings grant. Adds the two symptoms that follow from the installing account: a connection missing from the picker because it lacks the grant, and meetings cut off at 40 minutes because the app was installed from a Basic Zoom account.
…te the thread-event list Verification pass against develop. The Meet sign-in also asks for the name and email address that label the connection, adding the Meetings grant can ask for a reconnect, and the room thread also replies on unacknowledge and unassign.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Add a Traces guides section with the Cloudflare trace export guide
Refresh Assistant, Connections, and Workspaces screenshots with renamed demo data
…ance volume Brings the agent database-monitoring rewrites onto develop: per-engine coverage tables with each engine's setup steps, credentials modes, editing from the agent's Databases tab, and Advanced keys to reduce query performance events. Co-Authored-By: Ansh Garhewal <anshdevsin@gmail.com>
Co-Authored-By: Ansh Garhewal <anshdevsin@gmail.com>
Co-Authored-By: Ansh Garhewal <anshdevsin@gmail.com>
…ngine Co-Authored-By: Ansh Garhewal <anshdevsin@gmail.com>
…etrics Co-Authored-By: Ansh Garhewal <anshdevsin@gmail.com>
Co-Authored-By: Ansh Garhewal <anshdevsin@gmail.com>
Documents the km-kube-agent Helm values for container resources and priorityClassName (kloudmate/km-agent#146). Co-Authored-By: Ansh Garhewal <anshdevsin@gmail.com>
…ample Co-Authored-By: Ansh Garhewal <anshdevsin@gmail.com>
docs(agent): set resources and priority class for Kubernetes agent pods
database docs: per-engine setup, editing databases, and query performance volume
…I token and a service account key with its access, connecting it, and what each Degraded message means - New platform/settings/connections/atlassian: the admin setting that allows API tokens, the agent-interface scopes for reading and writing Jira and Confluence plus read:me and read:account for the site check, creating a scoped personal token (Atlassian MCP V2 app) and a service account key, the access a service account needs in projects and spaces, and the fix for each message the check reports. - Providers links the page from the Atlassian row; Manage connections says to point to Degraded to see why the last check failed.
eBPF monitoring sends a trace for every call, and those traces draw the service map. A caller that sends no telemetry still appears, by its Kubernetes workload or as incoming, and databases inside the cluster appear without Show external dependencies. Kubernetes workloads set to eBPF are traced too. The service map metrics list the label names they are stored and queried by.
…s which one when there are two, and Atlassian has one token field - Assistant settings: a section on one tool server per provider, the order (a server you connected before someone else's, then the oldest), the Not used by the Assistant label, and how to switch. - Atlassian links that section; Connect an account no longer offers Atlassian as an example of a choice under Authentication.
…removed coverages Co-Authored-By: Ansh Garhewal <anshdevsin@gmail.com>
docs(database-monitoring): fix the SQL Server agent-jobs grant, drop removed coverages
…and CLI uploads, how a frame finds its map, and each frame badge - New page at /rum/source-maps/, the URL the app's "Set up source maps" link opens. Options, flags, env vars and defaults are taken from @kloudmate/sourcemaps 0.1.2; matching, retention and badge causes from the backend symbolicator. - The Instrumentation Guide's Next list and the Web page's related resources link to it.
…ime, mapped stacks, Before the error, who was affected, retention The list explains which figures cover 24 hours and which 14 days, what New means and what search matches. Issue details explains what Users counts, Handled and Unhandled, which event the page shows, source-mapped stacks with a link to the source maps page, what Before the error records and keeps out, and which parts apply to RUM issues only. A new section says an issue is deleted after 15 days without an event and comes back as a new issue if the error recurs. Fresh screenshots replace the four old ones.
…ocker Drops the sample Spring Boot app and starts at attaching the OpenTelemetry Java agent, sending to the local KloudMate agent on localhost:4318. Co-Authored-By: Ansh Garhewal <anshdevsin@gmail.com>
docs(guides): Java zero-code instrumentation for a JAR, Tomcat, and Docker
… agent's config override or your own Collector Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Adds a section on the two HTTP rows that group requests without an endpoint label of their own, and corrects the drill-down's trace table, which now shows the request path and HTTP status instead of the span name.
…hydrating them Four pages under Logs → Cold Storage: an overview with roles and pricing, connecting a bucket (CloudFormation role, layout, export timing, coverage, replace/disable/delete), rehydrating a range (estimate, limits, statuses, Log Explorer, expiry, Glacier), and troubleshooting the verify and export errors. All four are draft: true until the feature ships to prod, and the five screenshots are placeholder images. Built and link-checked with the drafts off; anchors verified against dist.
The four Cold Storage pages move from Logs to Settings → Cold Storage, where the feature lives in the app, and are no longer drafts. The Settings overview lists them, and the Log Explorer page gains a Search Archived Logs section that points to them. The five screenshots are still placeholder images.
…ve tab, role step, and Rehydrations tab The AWS account ID in the Archive tab's role ARN is blurred. The estimate review and the Log Explorer reading a rehydration are still placeholders.
…estimate review, and the Log Explorer reading a rehydration
…uide, without the queue internals
… missing files get one too The estimate doesn't check storage classes, so Glacier hours still count in its rows and charge; they fail right away when loading reaches them and aren't billed. Troubleshooting gains the Glacier and missing-files reasons, and the destination failure email links to Settings → Cold Storage.
…by the Assistant badges Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Re-update MCP Integrations and Connections screenshots with Not used by the Assistant badges
…the client registration and token authentication the server supports
Co-Authored-By: Ansh Garhewal <anshdevsin@gmail.com>
…hropic Co-Authored-By: Ansh Garhewal <anshdevsin@gmail.com>
…upgrades Co-Authored-By: Ansh Garhewal <anshdevsin@gmail.com>
docs(agent): Upgrade the agent page
docs(guides): send Claude API and SDK traces to KloudMate
…nd managed-by notes
docs(gitops): fix Argo CD agent install, Kyverno webhook order, managed-by notes
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.