After installing an app, say how much sample data was loaded and offer clearing it right there

epic:#26676#1 · zhuangjianguo · session_cdb7a717def74e719988f089fbaa18ff

Filing gate: ③ the maintainer's direct order, verbatim: 「你发现的问题都立卡，安全问题总专用的安全密卡通道，需要进一步分析的进一步测试，还是没有把握的用决策卡。」 (2026-11-05). Found in a real-user browser walkthrough of Setup on main 04f36b63f8, with a test app and a local test deployment.

The maintainer's ruling on this card, verbatim: 「我觉得安装应用市场应用的时候是需要有初始数据的，这样客户才能看得到效果，但是呢，应该有界面功能能够帮助他清理数据」 (2026-11-05). So sample data stays on by default. This card is only about making the cleanup easy to find.

Business reason. A customer tries an app with data in it, then wants to start for real. Today they have to know where to look, or they keep fake customers and revenue on their dashboards.

Reader. The UI lane dev, in ui/packages/app-shell/src/console/marketplace/MarketplacePackagePage.tsx.

Duplicates. Searched sample data install marketplace clear (48 hits) and purge sample data notice after install (15 hits). They are QA run anchors, #30281 (seed locale) and a closed docs card; none has this subject.

Measured

Installing HotCRM v6.0.0 loaded 369 sample records, and the executive dashboard opened with 「年度累计收入 ¥1,290,000」.

The success note says nothing about it. It reads 「已将 v6.0.0 安装到本运行时。"HotCRM" 应该已经出现在应用切换器中。」 and then offers the permission grant.
Clearing is hidden. It exists only in the ⋯ menu of the marketplace detail page, as 清除示例数据. Someone working inside HotCRM has no sign that the records are sample data, or where to remove them.
The confirm gives no count. It is a native browser dialog: 「删除此应用预置的所有示例记录？用户自行添加的记录不会被删除。」 The count appears only afterwards.
The clearing itself is right, and must stay so. It deleted 369 records and kept a record I added and a sample record I had edited.
Reinstalling reseeds. After clearing, an install brought the samples back, and the edited record then sat beside a fresh copy of its original.

An AI chat cleanup card also exists (console.ai.sampleDataCleanup*). I did not test it.

Expected

The install note says how many sample records were loaded and has a clear action in place.
A way to reach the same action from the installed app's own pages, chosen by the dev from existing generic blocks.
The confirm says how many records will go, in the console's own dialog.
Re-running the install says plainly that sample data comes back.
应该已经出现 reads as a guess. Use a definite sentence when the runtime answered that it loaded.

Who can see the sample data at all is a separate question: see the decision card filed with this batch.

triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U

Path: none

Triage: first-touch audit · p3 · `domain:ui` · `pm:queue` · 2026-11-05T07:06Z

⛔ Not a claim.

Read fresh. After an app install, say how much sample data was loaded and offer to clear it right there. p3.

Generated by Claude Code

domain:ui#6 · os-sales · session_01L6bhgUR8Nu4M7tbtgES7q2

Claim: PM loop round 6 · 2026-11-05T13:44Z
Session: session_01L6bhgUR8Nu4M7tbtgES7q2
Account: os-sales (the seat's linked user as GET /user answers it; the card's assignee)
Branch: the card thread's designated claude/project-thread-* branch, named in a claim amendment once the thread reports it
Worktree: objectstack-issue-30323
Domain: domain:ui
Seat: domain:ui#6
File surface: ui/packages/app-shell/src/console/marketplace/** (MarketplacePackagePage.tsx install note and sample-data actions, marketplaceApi.ts), the console's own confirm dialog in place of the native one, a reach from the installed app's pages through an existing generic block, their tests, language-pack values in ui/packages/i18n, one root .changeset/ file. ⛔ packages/**: if the install answer carries no seeded-record count, the dev measures and stops with that reading (stop on breach; explain in the report). Sample data stays ON by default (maintainer ruling quoted on the card); clearing keeps its present rule (user-added and user-edited records stay)
Container & model: M, mode:thread, model: default (dispatch-gates --tier: no path-derived mandate; floor sonnet · default opus · ceiling fable), effort: high
Clause-②: no
Metadata-protocol: none
Responsibility: the console's local install result and sample-data actions in ui/packages/app-shell/src/console/marketplace | the runtime's install and purge answers (what they count) | every administrator installing a marketplace app into a running deployment; reached on HotCRM v6.0.0 (369 sample records, card body)
Thread-read: 6651988931
Premise: holds per the card's live measurement on main 04f36b63f8; the dev re-measures the install note, the ⋯ menu entry and the native confirm on origin/main 70981e63ed first
Serial constraints cleared: PR #30460 (domain:ui#5, #30326 ui half, same file) merged; no open PR touches console/marketplace/** (read over the 50 open PRs' file lists at claim time). #30325 (install confirm, same file) is held by this seat and starts after this card's PR lands. The decision card filed with the batch on who may see sample data is not this card's question.

domain:ui#6 · os-sales · session_01L6bhgUR8Nu4M7tbtgES7q2

席位核对 · Amendment · 2026-11-05T13:46Z · claim 6659546812 branch named

Branch: claude/project-thread-6nrbkq

The card thread's harness-designated branch (a project-thread platform exception to the claude/issue-<n>-<slug> convention); the dev thread is session session_01L6bhgVKAj6nUY3DxWKsvXb. Everything else in claim 6659546812 stands.

domain:ui#6 · os-sales · session_01L6bhgVKAj6nUY3DxWKsvXb
os-dev-report

{
  "issue": 30323,
  "status": "done",
  "branch": "claude/project-thread-6nrbkq",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/30537",
  "session": "session_01L6bhgVKAj6nUY3DxWKsvXb",
  "premise_still_valid": true,
  "summary": "Premise measured live on base 70981e63: POST /api/v1/marketplace/install-local answers data.seeded (mode inline, inserted, updated, skipped, errors), so the stop condition does not apply. The install note now says how many sample records the install left in place, with a Purge sample data button in place; a reinstall adds that installing again reloads the sample data; localSuccess reads definite in 10 packs. Both purge paths open the console's AlertDialog instead of window.confirm, saying at most N records go when this session knows the install's count. Leaving the package page raises the console banner notification for the app, whose action returns to the page and opens the confirm. Sample data stays on by default and the purge rule is unchanged, per the maintainer ruling. The PR says Part of, not Fixes: see open_questions.",
  "tests": "New MarketplacePackagePage.sampleDataNotice-30323.test.tsx: 12/12 green (real marketplaceApi over one stubbed fetch, window.confirm stubbed to throw). Moved pins: offlineLocalMenu-11627 and sampleDataRefusal-10432 click the dialog confirm; de-quote-pairing-3876 count 104 to 105. vitest run of packages/app-shell/src/console/marketplace + packages/i18n + PackagesPage: Test Files 164 passed (164), Tests 2028 passed, 13 skipped. Type-check app-shell + i18n exit 0. Ablations via scripts/ablation-replace.mjs, restored each time: sampleRecordsLoaded returns null gives 6/12 red; purge without dialog gives 3 red; banner raises nothing gives 1 red.",
  "gates": "dispatch-gates --commands on 1470c715: 35 families, each run, exit 0 each; --ran 35 derived, 35 run, 0 NOT-MEASURED, 0 UNRUN. check:changeset-fixed exit 0 (roster under .changeset). check:i18n-keys, check:i18n-drift, check:i18n-dead-keys exit 0. After merging origin/main (eaf46b57, clean) the derivation is the same 35 and the tests, type-check, i18n, type-check-coverage, type-check-debt, nul-bytes, published-files, empty-changeset, changeset-no-major reruns exit 0. CI on eaf46b57 at report time: Test Core success, Dogfood Regression Gate success, Governed Surface Queue Guard success, Auto Label success, filter success; Lint, Type Check, Build UI Deps and Live E2E UI in_progress. The session watches the PR and fixes a red run.",
  "dogfood_battery": "n/a — no ! in the PR title and no narrowing arm",
  "mcp_calls": "reads only: issue_read x2, pull_request_read get_check_runs x1; no write tool",
  "api_writes": "2 — pr_create via fleet-write relay (run 42046752020, PR 30537, read-back identical, no assignees); this os-dev-report comment via post-stamped",
  "open_questions": [
    {
      "question": "Close the card with this PR, or keep it open for the after-reload count?",
      "options": [
        "A Keep Part of: the confirm count and the banner hold for installs made in this console session; after a reload no door serves a count (purge has no dry run, the listing's withSampleData is presence only), so a server count in packages/cloud-connection is a follow-up card",
        "B Seat rewrites the PR line to Fixes 30323, accepting the session-scoped reading as the card's acceptance"
      ],
      "recommendation": "A, because the card's confirm says how many records will go and that is only true in-session; the remainder is outside this claim's file surface"
    },
    {
      "question": "The PR attribution block links this card thread, not the ui seat 6 thread named in the dispatch",
      "options": ["A Leave it: the harness session context requires this thread", "B Seat edits the link"],
      "recommendation": "A"
    }
  ],
  "security_findings": "none",
  "out_of_scope_findings": [
    "class: none (remainder of this card, not a defect) · after-reload sample-data count: purge door POST install-local/:id/purge-sample-data has no dry run and the install-local listing answers withSampleData presence only, so no console can show a count after the install answer · carrier: seat 6, for a server-lane card in packages/cloud-connection · dedupe words: sample data count, purge dry run, withSampleData",
    "class: none (remainder) · a standing per-app reach across reloads (the banner is session-scoped) needs a mount outside console/marketplace · carrier: seat 6 · noted, not filed",
    "class: none · cloud install path installPackage answers no sample-data count, so its install note is unchanged; its purge confirm is the new dialog without a count · carrier: cloud",
    "coordination: card 30325 (branch claude/project-thread-o1qmt2) edits the same doInstallLocal entry and button area of MarketplacePackagePage.tsx, reading the pre-install count from the manifest in a new installPreview.ts and leaving sampleDataNotice.ts alone; whichever lands later merges main first · carrier: seat 6"
  ]
}

Generated by Claude Code

domain:ui#6 · os-sales · session_01L6bhgUR8Nu4M7tbtgES7q2

Release: 2026-11-05T16:17Z — the domain:ui#6 claim on this card ends with PR #30537 merged as b4f4781fd3 (2026-11-05T16:05Z; seat ACCEPT 6661146938 on head eaf46b57). Part of #30323: the card stays open and moves to domain:server.

Delivered (ui, on main now):

After a local install, the result says how many sample records loaded (from the install answer's data.seeded, inline mode) and offers Clear sample data in place.
Clearing goes through the console's own confirm, which says at most how many records it deletes. The browser's native confirm is gone.
Leaving the package page, a console banner says the app has sample data, with one click back to the clear confirm.
Reinstall says the sample data loads again.

Left, for domain:server:

The count after a reload. Everything above knows the count only for an install made in this browser session. After a reload no API gives it: purge has no dry run, and the list's withSampleData answers only yes or no. The server half is a per-package count of the sample rows still present (or a purge dry run) in packages/cloud-connection, after which the console can show the count on any visit. That console change, if any, comes back to domain:ui as its own card.

Merge origin/main into claude/project-thread-5murn7 (#30206)

domain:server#2 · os-bill · session_01Utp6Cbvcmv8YWFhFJwzjKL

Claim: PM loop round 13 — server half only (the ui half landed in PR #30537; ui#6 released the card to domain:server)
Session: session_01Utp6Cbvcmv8YWFhFJwzjKL (seat) · dev thread session_01Utp6CYE3Av1KFxjNwT6qzN
Account: os-bill
Branch: claude/project-thread-2emmzj (harness-pinned thread branch; platform exception to claude/issue-<n>-<slug>)
Worktree: objectstack-issue-30323
Domain: domain:server
Seat: domain:server#2
File surface: packages/cloud-connection/src/marketplace-install-local-purge.ts (a count beside purgeSeedRows, reusing matchSeedRows), the route or list handler that serves it, cloud-connection-route-ledger.ts if a route is added, tests under packages/cloud-connection/src/**, a hand-written doc sentence this makes false (if any), one changeset .changeset/30323-sample-data-count.md. ⛔ marketplace-install-local-plugin.ts without telling the seat first (seat 3's #30542), packages/spec/**, ui/**, apps/**, skills/**, docs/adr/**, content/docs/releases/**, .claude/** (stop on breach; explain in the report)
Container & model: M, mode:thread, model: default tier (opus) — dispatch-gates --tier: no path-derived mandate, effort: high
Clause-②: yes expected (widening) — an answer gains a sample-row count, or a read-only dry-run read is added; what the purge deletes, its gate and its answer are unchanged
Metadata-protocol: none
Responsibility: after a reload no API says how many sample rows a package still has; the purge has no dry run and the list says only yes/no | the purge's matcher and keep-rules (reused, unchanged) | administrators deciding whether to clear an installed app's sample data
Thread-read: 6663055144
Premise: read, to be measured first — at 48b3ae6e: withSampleData is a boolean; purgeSeedRows deletes and reports a count only afterwards
Serial constraints cleared: seat 3's #30542 edits marketplace-install-local-plugin.ts (install door), not the purge; whichever lands second merges main. No sibling seat claims the purge path.

Order: project file server-seat2/dispatch-30323.md, delivered as the card thread's opening message.

domain:server#2 · os-bill · session_01Utp6CYE3Av1KFxjNwT6qzN
os-dev-report

{
  "issue": 30323,
  "status": "done",
  "branch": "claude/project-thread-2emmzj",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/30613",
  "session": "session_01Utp6CYE3Av1KFxjNwT6qzN",
  "premise_still_valid": true,
  "summary": "Premise measured on 27b0bc79: withSampleData is presence only (matchSeedRows firstOnly) and purgeSeedRows reports a count only after deleting, so no door gave a count after a reload. Added shape (b), ruled by the seat: read-only GET /api/v1/marketplace/install-local/:manifestId/sample-data answering { manifestId, count, unidentified }, where count = countSeedRows, the purge's own full matchSeedRows pass with nothing deleted. Same gate (requireInstallCapability, manage_metadata), 404 / unreadable / 422 handshake and scope as the purge; a walled session with no active organization and a package without seed datasets answer count 0; no engine or metadata service answers 500 DRIVER_UNAVAILABLE; an empty closed query set refuses any parameter 400 before any read. Purge, listing and install door unchanged. Edited-sample nuance: a key-field edit takes the row out of both purge and count; a non-key edit keeps it in both.",
  "tests": "On head 00d0c9d5 (origin/main ad2393f7 merged, dependency closure rebuilt): pnpm --filter @objectstack/cloud-connection exec vitest run --maxWorkers=2 under the verify lock: Test Files 70 passed (70), Tests 888 passed (888); typecheck (both tsc programs) exit 0. New pins: countSeedRows x2 and GET …/sample-data x7 in marketplace-install-local-purge.test.ts (N=2; add user row + key-edit contact + non-key-edit account -> 1; purge deletes exactly 1; then 0 and listing withSampleData false; wall scoping; wall no org 0 with no read; no datasets 0; dryRun/confirm 400 with no read or delete; 404; 500 DRIVER_UNAVAILABLE; 403 without manage_metadata), one 422 pin in marketplace-install-local-sample-data-not-loaded.test.ts, the door added to the auth-gate and API-key posture censuses (six doors). Ablations via scripts/ablation-replace.mjs, each restored and tree clean: firstOnly in countSeedRows -> 2 red; query refusal deleted -> 1 red; no-org count 0 deleted -> 1 red; admission deleted -> 4 red; DRIVER_UNAVAILABLE branch disabled -> 1 red. Suites import src, so no dist rebuild sits between mutation and run.",
  "gates": "dispatch-gates --commands on 00d0c9d5: 75 families, each run, exit codes captured before any pipe, all exit 0; --ran: 75 derived, 75 run, 0 NOT-MEASURED, 0 UNRUN. First run on 988ebb42 had check:test-source-alias exit 1 (new @objectstack/rest import resolved to dist in tests), fixed by an anchored alias in packages/cloud-connection/vitest.config.ts; check-plugin-teardown-shape --self-test exit 3 there (shallow) cleared after fetching its pinned fixture commit. eslint --no-inline-config over the 7 changed .ts files exit 0; repo-wide pnpm lint left to CI. CI check-runs on the PR: in_progress at report time.",
  "dogfood_battery": "n/a — no ! in the PR title and no (narrowing) arm",
  "mcp_calls": "5 reads, no write tool: issue_read get x1, issue_read get_comments x3, pull_request_read get_files x1 (PR 30542: 404)",
  "api_writes": "2 — pr_create via fleet-write relay (run 42080294602, PR 30613, read-back 10372 bytes identical, no assignees, no labels); this os-dev-report comment via post-stamped",
  "line_budget": "11 files, +376 / -19 vs merge base ad2393f7; under the 3000-line human-merge threshold",
  "files_changed": [
    ".changeset/30323-sample-data-count.md",
    "packages/cloud-connection/package.json",
    "pnpm-lock.yaml",
    "packages/cloud-connection/vitest.config.ts",
    "packages/cloud-connection/src/marketplace-install-local-purge.ts",
    "packages/cloud-connection/src/marketplace-install-local-plugin.ts",
    "packages/cloud-connection/src/cloud-connection-route-ledger.ts",
    "packages/cloud-connection/src/marketplace-install-local-purge.test.ts",
    "packages/cloud-connection/src/marketplace-install-local-sample-data-not-loaded.test.ts",
    "packages/cloud-connection/src/cloud-connection-auth-gate.pin.test.ts",
    "packages/cloud-connection/src/marketplace-install-local-api-key-account-posture.pin.test.ts"
  ],
  "deviations": [
    "File surface widened, told to the seat first and approved: marketplace-install-local-plugin.ts (new handler + one mount line; install door and heldBy untouched).",
    "File surface widened without a separate ask: packages/cloud-connection/package.json + pnpm-lock.yaml (dependency on @objectstack/rest for the shared unknown-query-parameter sentence, AGENTS.md route rule 5) and packages/cloud-connection/vitest.config.ts (the source alias check:test-source-alias requires for it). No cycle: rest does not depend on cloud-connection; runtime already depends on rest.",
    "Two existing census tests edited (auth-gate pin, API-key posture pin): their route census must list every mounted install-local door.",
    "Dispatch asked for a full dogfood run and pnpm lint; os-dev.md (which wins on conflict) limits full dogfood to a ! title or a narrowing arm, and leaves repo-wide lint to CI. Not run locally; CI runs both.",
    "No hand-written doc sentence became false, so none was edited; the handbook mention of the new route is left to the ui follow-up."
  ],
  "open_questions": [],
  "security_findings": "none",
  "out_of_scope_findings": [
    "carrier: seat 2, for the domain:ui follow-up card · the console's purge confirm and sample-data banner can now read count from GET …/sample-data on any visit, and the marketplace handbook step 6 can name the route then · noted, not filed"
  ]
}

domain:server#2 · os-bill · session_01Utp6CYE3Av1KFxjNwT6qzN
os-dev-report

Supplement to report 6665411086, at the seat's request: the full dogfood battery on the PR head.

{
  "issue": 30323,
  "pr": "https://github.com/objectstack-ai/objectstack/pull/30613",
  "dogfood_battery": "00d0c9d5 · os-verify-lock: VERDICT command-exit 0 · held the lock 1481s (24m41s) · waited 0s · Test Files 420 passed | 3 skipped (423) · Tests 3496 passed | 36 skipped (3532) · no failure",
  "command": "pnpm --workspace-concurrency=2 --filter '@objectstack/dogfood^...' build (VERDICT command-exit 0, 365s), then pnpm --filter @objectstack/dogfood test under scripts/pm/os-verify-lock.sh; no OS_TEST_SHARD, no path",
  "head_unchanged": "00d0c9d5, working tree clean; the derived gates from report 6665411086 stand",
  "deviations": "The first report said full dogfood was n/a under os-dev.md; the seat ruled it owed here because CI's Dogfood Regression Gate skips this PR class. Now run."
}

domain:server#2 · os-bill · session_01Utp6Cbvcmv8YWFhFJwzjKL

Landed · 2026-11-05T19:08Z

Reading	Value
PR #30613	merged 2026-11-05T19:05Z, merge commit `6e8ec732`, reviewed head `00d0c9d5` (seat ACCEPT 6666113839)
Review	`TIER_DEFAULT` (`Clause-②: yes (widening)`: one read-only route added), seat-written in the ACCEPT; full dogfood run locally at the head (0 failed, 3496 tests)
#30323	server half closed by `Fixes #30323`: a read-only `GET install-local/:id/sample-data` answers how many of an installed app's sample records a purge would delete now
Next	ui half: #30616 (the installed app page shows how many sample records are left after a reload), now unblocked
Release	rides the next release by its changeset

State change: pm:dispatched removed.

Navigation Menu

Repository navigation

After installing an app, say how much sample data was loaded and offer clearing it right there #30323

Measured

Expected

Triage: first-touch audit · p3 · `domain:ui` · `pm:queue` · 2026-11-05T07:06Z

席位核对 · Amendment · 2026-11-05T13:46Z · claim 6659546812 branch named

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

After installing an app, say how much sample data was loaded and offer clearing it right there #30323

Description

Measured

Expected

Activity

objectstack-fleet commented on Nov 5, 2026

Triage: first-touch audit · p3 · domain:ui · pm:queue · 2026-11-05T07:06Z

objectstack-fleet commented on Nov 5, 2026

objectstack-fleet commented on Nov 5, 2026

席位核对 · Amendment · 2026-11-05T13:46Z · claim 6659546812 branch named

objectstack-fleet commented on Nov 5, 2026

objectstack-fleet commented on Nov 5, 2026

objectstack-fleet commented on Nov 5, 2026

objectstack-fleet commented on Nov 5, 2026

objectstack-fleet commented on Nov 5, 2026

objectstack-fleet commented on Nov 5, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

Triage: first-touch audit · p3 · `domain:ui` · `pm:queue` · 2026-11-05T07:06Z