Tasks completed: 1/1
- Port auth key normalization tests (commit #177, MISC-24)

SUMMARY: .planning/phases/19-misc-fixes-cleanup/19-02-SUMMARY.md

…anager

… comms discipline

Phase 4 adds:
- DELEGATE dispatch verification + attachment passthrough (openclaw#177)
- Time-based self-continuation with agent-specified delay (openclaw#176)
- Pre/post-compaction lifecycle hooks (openclaw#178)
- Config hot-reload gap (openclaw#182)
- Context-pressure canary confirmation (smaller model per figs)

Updated issue tracking: openclaw#174 closed, openclaw#175 root cause found.
Added Discord comms discipline section (Storm Lag Protocol + figs directives).
Thomistic basis: Summa II-II Q47-Q53, prudentia survives forgetfulness via voluntas.

Add +Ns delay suffix to [[CONTINUE_DELEGATE: task +30s]] bracket syntax.
When present, spawnSubagentDirect is called after a setTimeout delay,
bounded by minDelayMs/maxDelayMs from continuation config.

Marker event fires immediately so parent session knows delegation is
pending. Timer does not survive gateway restart (v1 limitation, see openclaw#176).

Changes:
- tokens.ts: extend ContinuationSignal delegate variant with delayMs,
  parse +Ns suffix from bracket body
- agent-runner.ts: extract doSpawn closure, wrap in setTimeout when
  delayMs > 0, fire delegate-pending event immediately
- tokens.test.ts: 7 new tests for timed delegate parsing

57/57 token tests pass.

… 136, debug logging assigned to Silas

* feat(settings): add autostart toggle in advanced settings

Enable users to configure autostart behavior through a new toggle in the advanced settings UI. The backend now checks user preferences on startup to enable or disable autostart accordingly, with the setting persisted and manageable via a new Tauri command. This provides better control over app initialization without affecting existing functionality.

* immediate toggle effect?

---------

Co-authored-by: CJ Pais <[email protected]>

* Initial plan

* fix(browser): 追踪原始端口映射以处理 EADDRINUSE 回退

移植上游 commit 8e55503 的修复：

**问题**：
当请求的端口已被占用时（EADDRINUSE），Node.js 的 server.listen()
可能会自动分配一个不同的端口。但代码在 serversByPort Map 中使用
实际分配的端口作为键存储服务器，却在查找时使用原始请求的端口，
导致后续调用无法找到已存在的服务器实例。

**修复**：
- 保存 originalPort（原始请求的端口）
- 使用 originalPort 作为 serversByPort Map 的键
- 这样即使实际端口改变，也能正确查找和清理服务器实例

**验证**：
- ✅ 构建通过
- ✅ 相关测试全部通过 (3/3)
- ✅ 未修改用户可见文本（CDP 协议内部错误保持英文）

Co-authored-by: jiulingyun <[email protected]>

---------

Co-authored-by: copilot-swe-agent[bot] <[email protected]>
Co-authored-by: jiulingyun <[email protected]>

* feat(data-explorer): searchable icon-grid view of ingested connector data (openclaw#174)

Closes openclaw#173

- Backend: GET /v1/connectors/data — paginated, filterable, searchable
  query over connector_data_index (one entry per source doc). Adds
  store.list_connector_index_entries() with regex text search, provider
  filter, and skip/limit pagination.
- Frontend hook: useConnectorData(q, source, page) with abort-controlled
  fetch and debounce handled in the page.
- Frontend page: /dashboard/data — Google Drive-style icon grid with
  per-provider icons (Slack, Notion, Drive, Fathom), debounced search
  bar, source filter pills, spring-animated detail drawer, pagination,
  and empty state.
- Header: adds "Data" nav link with FolderSearch icon.

Co-authored-by: Claude Sonnet 4.6 <[email protected]>

* checkpoint (openclaw#177)

* feat(memory): add admin debug endpoints

* feat(memory): wire chat and slack publishers

* feat(memory): make synthesis profiles pluggable

* feat(frontend): add memory admin debug page

* feat(backend): publish refresh outputs to memory

* chore(dev): start memory with root dev loop

* feat(memory): add postgres storage provider

* feat(connectors): add slack memory sync observability

* docs(memory): document technical orchestration

* build(memory): generate and verify code mind map

* fix(frontend): remove leftover connectors merge marker

* docs(engram-chat): add connector migration matrix

* feat(engram-chat): migrate connector oauth parity

* feat(engram-chat): migrate connector sync to memory

* feat(engram-chat): migrate fathom webhook ingest

* docs(engram-chat): record remaining backend legacy surface

* chore(repo): remove stray ds store files

* refactor(routes): retire legacy backend chat surface

* fix(engram-chat): expose connector sync state collection

* feat(memory): include references in chat retrieval

* feat(memory): tighten connector retrieval semantics

* feat(engram-chat): add mongo-backed chat prompt configs

* feat(engram-chat): compose chat prompts at runtime

* docs(engram-chat): document prompt override workflow

* fix(engram-chat): restrict prompt overrides to admins

* feat(frontend): add prompt admin controls to memory page

* feat(prompts): scope product prompt control to admins

* build(dev): add oauth tunnel startup command

* docs(engram-chat): document stable oauth callbacks

* feat(engram-chat): add direct chat fallback for local runtime

* feat(frontend): enrich memory admin prompt controls

* feat(frontend): add admin workspace debug portal

* feat(frontend): fold memory into brand brain experience

* fix(memory): repair admin event and stats endpoints

* fix(frontend): expose full prompt catalog and route to chat

* feat(memory): add user-scoped memory steering

* feat(frontend): add explicit brand team switcher

* refactor(frontend): move prompt governance into admin portal

* feat(frontend): streamline brand switcher and add new brand entry

* fix(frontend): route team workspace to backend api

* build(dev): mirror runtime logs to root file

* fix(frontend): always open brand switcher menu

* refactor(frontend): remove team and data from primary nav

* feat(frontend): embed data explorer into connectors

* feat(frontend): add connect and explore tabs to connectors

* fix(frontend): load connector explorer data from backend

* fix(chat): keep product context across follow-up turns

* fix(chat): ground brand-scoped turns in profile context

* fix(chat): add website identity to brand context

* docs(traceability): capture aligned scope

* feat(chat): trace sync runs and expose admin inspection

* test(chat): add context regression harness

* feat(chat): expose retrieval provenance in traces

* feat(chat): trace streamed chat runs

* feat(frontend): inspect chat traces from admin portal

* feat(multi-brand): full multi-brand support — routing, access control, delete, refresh fix (openclaw#180)

* feat(multi-brand): support multiple brands per user (openclaw#178)

Backend:
- Store: add list_brand_profiles_for_user (all owned+member brands),
  create_brand_profile (new insert), upsert_brand_profile_by_id (owner-only update by id)
- Brand router: POST /v1/brand creates a new brand; GET /accessible now returns
  ALL owned and member brands via list_brand_profiles_for_user
- Add _get_brand_for_user helper for consistent owner/member access control
- Thread brand_id query param through /profile, /website-identity, /refresh,
  /refresh-status, /today, _trigger_refresh_task, _generate_daily_brief_task
- POST /profile accepts ?brand_id= to update a specific brand (owner only)
- Chat: ChatRequest.brand_id optional field; _get_context and run_content_pipeline
  tool respect active brand_id

Frontend:
- BrandContext: add createBrand(); saveTracking and setBrandThemeEnabled now
  pass ?brand_id= when activeBrandId is set
- useDailyBrief: accept brandId option, thread through today/refresh/refresh-status/
  website-identity API calls
- Header BrandSwitcher: "New brand" inline form to create additional brands

Co-Authored-By: Claude Sonnet 4.6 <[email protected]>

* fix(multi-brand): clear stale activeBrandId on 403/404 profile fetch

If localStorage has a brand_id the user no longer has access to,
the profile request returns 403. Clear the key and retry without it.

Co-Authored-By: Claude Sonnet 4.6 <[email protected]>

* feat(header): always show brand dropdown with + Add brand button

Removed the single-brand shortcut that rendered a plain link with no
dropdown — the trigger now always opens the dropdown, giving every user
access to the brand list and the inline "Add brand" form.

Co-Authored-By: Claude Sonnet 4.6 <[email protected]>

* feat(multi-brand): add brand via website URL, not brand name

POST /v1/brand now accepts website_url, derives a seed name from the
domain, persists the new brand, then fires _trigger_refresh_task in the
background — same extraction flow as first-brand onboarding.

Frontend: createBrand() accepts a URL; Header "Add brand" input shows a
Globe icon with placeholder "example.com" and a "Go" submit button.

Co-Authored-By: Claude Sonnet 4.6 <[email protected]>

* chore: restore stashed session changes (brand delete, context ACL, team brand_id, store methods)

Co-Authored-By: Claude Sonnet 4.6 <[email protected]>

* fix(multi-brand): correct brand routing across all pages and fix refresh crash

- store: add get_brand_for_user() — single access-control resolver (owner or member)
- brand.py: remove local _get_brand_for_user(), replace all 7 call sites with store method
- chat.py: replace two inline access-control blocks with store.get_brand_for_user()
- context.py: simplify _assert_brand_access() to use store.get_brand_for_user()
- brand.py _synthesize_brief: fix NameError — bare brand_id → str(brand.id or "")
- useDailyBrief: add brandId to useCallback dep arrays (fetchBrief, fetchWebsiteIdentity, fetchRefreshStatus) — was stale on brand switch
- today/page: pass activeBrandId to useDailyBrief
- brand-brain/page: pass brand_id param to /v1/brand/profile and /v1/brand/website-identity fetches
- team/page: add useBrandProfile, pass brand_id to all 4 team API calls, add brandParam to fetchTeam deps

Co-Authored-By: Claude Sonnet 4.6 <[email protected]>

---------

Co-authored-by: Claude Sonnet 4.6 <[email protected]>

* feat(backend): add agent web research endpoint

* feat(chat): trigger web research from search intent

* feat(chat): add search policy prompt section

* feat(ux): show loading indicator when a new brand is being initialized

- toast.loading/success/error pair in BrandSwitcher for full-lifecycle feedback
- animated skeleton loading row appears in brand list while creation is in progress
- all brand-switch and delete buttons disabled while creating to prevent race conditions
- deferred setActiveBrandId until after fetchBrand resolves so the switcher does not flip to the new brand before its profile is ready

Closes openclaw#181

Co-Authored-By: Claude Sonnet 4.6 <[email protected]>

* fix(frontend): surface saved search data in explore

* feat(chat): stream agent activity summaries

* fix(frontend): soften technical thinking summaries

* feat(chat): add langfuse tracing support

* feat(admin): surface langfuse metadata in traces

* speed up brand refresh

* fix(openclaw): bind gateway to 0.0.0.0 on Render

Render requires the service to listen on 0.0.0.0 for port detection.
Add dockerCommand to override the Dockerfile CMD with --bind lan --port 8080,
and remove the no-op OPENCLAW_GATEWAY_BIND env var (not read by the codebase).

Co-Authored-By: Claude Sonnet 4.6 <[email protected]>

---------

Co-authored-by: Claude Sonnet 4.6 <[email protected]>
Co-authored-by: sdntsng <[email protected]>