Skip to content

feat: add member task snapshot template - #10175

Closed
Zedmor wants to merge 3 commits into
mainfrom
feat/member-task-board
Closed

Zedmor wants to merge 3 commits into
mainfrom
feat/member-task-board

Conversation

@Zedmor

@Zedmor Zedmor commented Sep 11, 2026 •

Copy link
Copy Markdown
Contributor

Problem / Motivation

Crew Members need a clear way to show task progress while owners create work and give instructions in the member’s existing chat.

Depends on #8468. This PR targets feat/crew-webview for review. Merge it only after the dashboard framework lands and this PR is retargeted to main; full CI runs only for PRs targeting main.

Why it matters

Owners should see the same task facts that their member uses. A worker saying it is done must not look like an accepted result, and failed dispatch preparation must leave the task available to retry.

What changed (motivation → approach → change)

The bundled Tasks template shows a published ledger snapshot in Crew summary → Dashboard. Owners create tasks, steer work and give acceptance decisions in the existing member chat. To do, In progress, Blocked, Review, Done and Closed come from ledger fields. Worker completion reaches Review; acceptance reaches Done. Cards show reports, criteria, decisions, verification results and evidence. Labels and dates are translated, and the document renders text inside the existing sandbox.

spawn_run(work_item_id=...) connects one task to a worker in the caller’s own ledger. The gateway prepares the worker and binds its generated identity before the first stream. Scoped dispatch defaults to kirocrew-worker and checks both reporting tools using the allocated provider’s actual backend and directory. Explicit tool exclusions override a broad allowlist. KAS uses its user-level spec and wildcard projection; the other supported backends use their project-aware resolver. The worker gets its own runtime identity, ordinary approval and inherited private memory. Briefs and reports require the binding and item to name the same worker; an interrupted bind cannot authorize an obsolete worker. Failed preparation leaves the task retryable. Member DMs receive work and panel tools without new approval grants. Their tracked-task procedure uses scoped spawn and a bounded patrol to check reports, steer stalls and publish updates.

panel_publish(template="tasks", data={summary?}) reads the verified caller’s own ledger. Only the optional summary comes from model-supplied data. The snapshot omits event history and worker session keys; an assignment flag is enough for its lanes. Publication refreshes the existing panel query. The document shows its publication time, missing or truncated data, and the chat path for changes or updates. The shared publication scrubber retains field-name context before persistence, removing named credentials while preserving surrounding prose and structure. This template replaces the proposed native member task table for the first release.

flowchart LR
  A[Member chat]:::ctx --> B[Work ledger]:::changed
  B --> C[Scoped worker]:::added
  C -->|reports| B
  B -->|gateway reads on publish| D[Tasks template]:::added
  classDef added fill:#DCFCE7,stroke:#16A34A,color:#14532D,stroke-width:2px
  classDef changed fill:#FEF3C7,stroke:#D97706,color:#78350F,stroke-width:2px
  classDef removed fill:#FEE2E2,stroke:#DC2626,color:#7F1D1D,stroke-dasharray:4 3
  classDef ctx fill:#E0F2FE,stroke:#0284C7,color:#0C4A6E
  linkStyle 1,2,3 stroke:#16A34A,stroke-width:2px
Loading

🟩 added · 🟨 changed · 🟥 removed · 🟦 unchanged

Owners use the member chat to direct work and the published board to see its recorded progress.

Tests

  • 701 focused backend tests pass, with five expected failures. They cover private dispatch, verified worker identity, gateway-sourced task facts, named-artifact credential scrubbing, acceptance, borrowed-identity refusal, duplicate dispatch, approval races and retries after preparation failures. Four regression cases reproduce an interrupted or superseded binding on the prior code; the fix refuses both reads and reports while allowing the current worker to continue.
  • Project overrides without reporting tools fail before binding, and a valid custom agent can retry the same item. Exclusions of a required tool, its server or a matching wildcard refuse dispatch; unrelated exclusions remain valid across the four supported backends. Conflicting pool/worker directories and KAS user/project specs follow the allocated backend’s actual resolution. Storeless workers use the dedicated path and strict reporting identity.
  • The full website suite passed: 2,019 files and 32,148 tests, with one expected failure and two skips. The unchanged template and document builder have 234 passing focused tests. The Electron suite passed with its pinned local dependencies: 1,849 passes and one skip.
  • Production build, TypeScript, bundle-size, chunk-cycle and i18n render comparison against the parent passed. Python formatting, import order, lint and type checks, documentation, harness parity, memory boundaries, duplication and applicable static gates pass. All 19 i18n checks passed. Both extracted local review contracts were run with the available model; pinned-model parity is not claimed.
  • The full backend run completed with 97,193 passes, 167 failures and 378 errors on this host. The five feature-related failing expectations pass in the focused suites. Of the remaining results, 466 fail because this host lacks atomic no-replace rename; 73 reproduce on the exact parent commit. One daemon-start failure did not reproduce on either version. Full CI remains required after retargeting to main.

Manual verification

The real Members page, template, themes and i18n code were exercised with intercepted API responses. Browser verification covered the board, card evidence, sandbox grants, instructions sent through member chat, dark and light themes, task content at 390px and 320px without horizontal overflow, and empty, unavailable and truncated snapshots. The expanded card shows distinct Pull request, Evidence and Verification result labels; the board names chat as the place to accept results. The recording demonstrates this fixture-based flow. The screenshots were visually inspected; this is not a human usability study. The binding correction changes no UI code, so the attached captures still show the current interface.

Screenshots / video

Task board with progress, blocked work and acceptance

Member chat beside the published task summary

Chat instructions, light theme, narrow layout and empty state

Give task instructions in member chat

Light theme

390px task board

Empty task snapshot

320px task board

Unavailable snapshot

Truncated snapshot

task-template-flow.webm

Related Issues

No linked issue: requested directly to create tasks, steer the member and monitor progress, with the first release using a template and existing chat.

Checklist

  • One commit, with a Conventional Commits title.
  • All required CI passes on main — pending feat(members): give every crew a configurable webview in its drawer #8468 and retargeting; local full-backend limitations are listed above.
  • Focused tests pass and cover the new behavior.
  • Self-review completed; both local review contracts run with the available model. Pinned-model parity is not claimed.
  • Documentation updated.
  • No secrets, credentials, or internal references in the diff.

chenmingwei23 and others added 2 commits September 10, 2026 02:19
A crew that runs unattended accumulates the state an operator actually wants:
how many workers it is holding, which one is stuck, what it is waiting on. The
drawer showed activity counts, a path list and config instead, none of which
answer those. This adds a free-form HTML webview that the crew fills in at
runtime.

The split that makes free-form HTML safe: a human authors the TEMPLATE --
versioned in the repository, or dropped on disk by the operator -- and the crew
publishes only DATA. The template directory is fenced from agent file tools, so
a crew cannot author its own markup; the only way it influences its webview is
by publishing a data object. The data lands as an inert application/json island,
escaped so a published string cannot close the script element early, and a
template reads it with JSON.parse and renders through textContent only. The
composed document is served behind a single-use minted URL into an iframe
sandboxed to allow-scripts and nothing else.

A crew whose name matches an installed template gets that template; a crew with
no template of its own falls back to a generic one that renders any data object.
That is the entire wiring mechanism -- installing <crew-name>.html is the act of
wiring a bespoke view up, so there is no registry, no mapping table and no
per-crew config to keep in sync.

The MCP server ships registered but gated off for every agent (opt_in), so a
default session pays nothing for the capability: not a tool-list entry, not a
line of context, not a spawned backend. An agent that should publish a panel is
granted the set in its own spec, and a refresh keeps an existing grant current
without ever minting one.

The publishing crew is derived from the vetted calling session and never from the
request body, so one crew cannot publish a webview that presents as another's. A
body field naming a crew is rejected by the schema rather than ignored, so a
caller is never left believing it retargeted the write. The drawer's read at GET
/api/members/{slug}/panel is cookie-authed; the /api/agent-panel writes sit
behind the MCP-only prefix.

Published field order is preserved end to end. Field order is presentation -- a
template renders a stat strip in key order -- so the record is dumped without
sorting keys instead of silently alphabetising the operator's dashboard.
# Conflicts:
#	docs/feature-map/README.md
#	src/kiro_crew/sandbox.py
#	src/kiro_crew/security/paths.py
#	website/src/i18n/locales/bn.json
#	website/src/i18n/locales/de.json
#	website/src/i18n/locales/es.json
#	website/src/i18n/locales/fr.json
#	website/src/i18n/locales/hi.json
#	website/src/i18n/locales/it.json
#	website/src/i18n/locales/ja.json
#	website/src/i18n/locales/ko.json
#	website/src/i18n/locales/pt.json
#	website/src/i18n/locales/ru.json
#	website/src/i18n/locales/zh-CN.json
@Zedmor
Zedmor requested review from a team and hoang-phan98 September 11, 2026 19:31
@Zedmor

Zedmor commented Sep 11, 2026 •

Copy link
Copy Markdown
Contributor Author

Intent: Let an owner create tasks and steer a Crew Member in the existing member chat, and monitor a published task snapshot through the Tasks template in Crew summary's dashboard. Connect existing scoped worker dispatch to the real ledger while preserving private memory and ordinary tool approval.

Not a goal: A separate task database, native task action controls, archive storage, or a continuously polled board.

User-directed scope change: after comparing this feature with #8468, the owner explicitly selected "Template + existing chat (recommended)" for the first release. That choice replaces the earlier native Tasks tab and archive intent. This PR builds on #8468 and must be retargeted to main after that framework lands.

@github-actions github-actions Bot added the readiness: checking Automated validation is still running label Sep 11, 2026
@github-actions

github-actions Bot commented Sep 11, 2026 •

Copy link
Copy Markdown
Contributor

Design Review (Fable 5) — 🟡 CONCERNS

Design-level review of 3e518c04055b924ff847269255b98e88cfe04ffe — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

Design-Verdict: CONCERNS

Sound gateway-sourced snapshot and verified-binding design; the new tool-exposure preflight is a parallel resolver that can silently drift from real backend semantics.

Watch

  • agent_exposes_mcp_tools (src/kiro_crew/agent_sdk/drivers/acp.py) reimplements each backend's exposure semantics — private imports of kas_agents._project_tools and tool_aliases._parse_tool_refs, plus new fnmatch logic for excludedTools/disabledTools, behind a hardcoded four-backend switch that returns False for anything else. When a backend's projection or exclusion semantics change (or a fifth backend lands), this preflight drifts: a false admit yields a worker that can never call work_report, so its item sits in "In progress" until the patrol escalates; a false refusal blocks scoped dispatch with work_reporting_unavailable. Both are recoverable, but the drift is silent because nothing ties this copy to the runtime's own projection.
    Clears when: a parity test exercises agent_exposes_mcp_tools against the same specs the live projection resolves (per backend), or the check delegates to the projection code the runtime itself runs.

Suggestions

  • The "kirocrew-worker" default is applied at three layers (messaging.py, admission.py, run.py); collapse to one owner so a future default change can't half-apply.
  • The _scrub prefix-splice ("prepend field: , redact, strip") changes redaction behavior for all panel templates and destroys the whole value when a match merely overlaps the prefix boundary; an explicit key-aware parameter on redact_credentials would express the same intent without the round-trip fragility.

[DESIGN-REVIEWED] 3e518c0

@github-actions

github-actions Bot commented Sep 11, 2026 •

Copy link
Copy Markdown
Contributor

UX Review (Fable 5) — 🟡 CONCERNS

UX-level review of 3e518c04055b924ff847269255b98e88cfe04ffe — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

All evidence is in hand. The change is a read-only "Tasks" kanban snapshot template rendered in the existing sandboxed member webview, with all strings localized across 13 locales; the blind read never ran because no screenshots were committed to the repository (the PR's captures live only as description attachments).

UX-Verdict: CONCERNS

Copy and states are solid, but zero screenshots were committed, so the blind read never ran — no cold-read evidence for the entire new Tasks board.

Watch

  • The "Review" lane is where the owner must act (the PR's stated core: worker-done must not read as accepted), yet unlike "Done"/"Closed" it carries no hint — a first-time user sees a completed-looking task parked in "Review" with only the distant top-of-board chat_hint explaining what to do. Hits on every completed task, persistent. Fix: add review_hint, e.g. "Awaiting your acceptance in chat."

Evidence gaps

  • Blind read not performed: no images under temp-screenshots/ or .github/screenshots/ — PR-description attachments are not committed artifacts. Commit the captures the description already shows (board, expanded card, empty, unavailable, truncated, light theme, 390px/320px).
  • Stale-task warning ("Awaiting an update") appears in no listed capture; commit one showing it.

Suggestions

  • memberTaskPanel.truncated is reused inside a card's Evidence list (>20 artifacts), where "Only part of this snapshot is shown" misstates the scope — add an evidence-specific string, e.g. "Only some evidence is shown."

[UX-REVIEWED] 3e518c0

@github-actions

github-actions Bot commented Sep 11, 2026 •

Copy link
Copy Markdown
Contributor

First Principles Review (Fable 5) — 🟡 CONCERNS

Premise-level review of 3e518c04055b924ff847269255b98e88cfe04ffe — why this exists and whether the shipped surface is the smallest honest version. Updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

First-Principles-Verdict: CONCERNS

The member's dispatch loop moves to spawn_run, but the two approval-free session_* write grants that loop was the sole stated justification for stay behind.

Not justified as shipped

  • 4. The member's standing instructions drop session_create/session_send/session_read_message — _MEMBER_DASHBOARD_GRANTS (src/kiro_crew/agent.py:5820) auto-approves session_send and session_stop because, in its own words, "the dispatch loop this feature exists for (create → seed → patrol → stop) stalls on an approval prompt at its second step with nobody at the keyboard." This diff deletes that loop; the new one (spawn_run/spawn_steer/spawn_list) takes ordinary approval. The grants and the now-false comment remain.
  • 7. Named-credential redaction for every panel template — symptom-level: the blindness is redact_credentials needing key/value context, and only the panel walker was taught to supply it.
  • 8. The pre-binding reporting-tools check — one consumer, generalized.

What this change ships

Inventory (10 of >10 items) — 7 justified

Intent: let an owner create tasks in the member's chat and watch their progress on a board. ADDITION.

  1. Bundled tasks board in Crew summary → Dashboard; worker done lands in Review, only acceptance reaches Done — justified
  2. spawn_run(work_item_id=…) dispatches one ledger item; the gateway binds the worker before its first turn — justified
  3. Member chat gains the work-ledger and panel tool sets, three MCP servers under one identity, no new auto-approvals — justified
  4. Member instructions swap session_* dispatch/patrol for ledger + scoped spawn + spawn_list — leaves the two session_* write grants without their stated cause
  5. Publishing tasks discards model-supplied conductor/items and reads the caller's own ledger; worker_session_key becomes has_worker — justified
  6. A worker whose item no longer names it is refused briefs and reports — justified
  7. Every panel template now redacts credentials named by their field, replacing the whole value — symptom-level
  8. Dispatch refused with work_reporting_unavailable when the resolved agent lacks both work tools — one consumer, generalized
  9. Panels can carry host display context (#kirocrew-context: labels, locale, publish time) — justified
  10. panel_publish's confirmation no longer counts top-level fields — undeclared, harm-free

Not listed: _slot_running recognizing subagent: keys, spawn.py joining REFLEXIVE_TOOL_MODULES, ledger workers always taking the dedicated runtime path, the RFC Phase 4 rescope, the capture-harness ThemeProvider.

Watch

  • Item 7 is a point patch. Grepping src/kiro_crew/**/*.py for recursive dict/list scrubbers that call redact_credentials per value returns 3; this fixes 1. session_summary.redact_payload (session_summary.py:391) and mcp_core._redact_json_strings (mcp_core.py:2378) still scrub each value without its key, and both persist before render.
    Clears when: the two siblings pass field context too, or the description states this fix covers panels only and names them as left.
  • Item 4's grants: the comment justifying them describes a loop the diff removes.
    Clears when: session_send/session_stop leave _MEMBER_DASHBOARD_GRANTS, or a named caller still needs them approval-free.

Subtractions

  • Drop "@kirocrew-dashboard/session_send" and "@kirocrew-dashboard/session_stop" from _MEMBER_DASHBOARD_GRANTS (src/kiro_crew/agent.py:5820) — back to _CONDUCTOR_DASHBOARD_GRANTS; the member prompt names no session_* verb after this diff.
  • Drop the server and tools parameters of agent_exposes_mcp_tools (agent_sdk/drivers/acp.py:52) — 1 consumer (subagent_manager/run.py:882) and it always passes mcp_work.SERVER_NAME + WORKER_TOOLS; take the singular agent_can_report_work(agent, *, work_dir, backend).

[FIRST-PRINCIPLES-REVIEWED] 3e518c0

@github-actions

github-actions Bot commented Sep 11, 2026 •

Copy link
Copy Markdown
Contributor

Opus 4.8 Review — ✅ no blocking findings

Reviewed 3e518c04055b924ff847269255b98e88cfe04ffe — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] 3e518c0

Verdict parsed from the review's SHA-scoped output markers for commit 3e518c04055b924ff847269255b98e88cfe04ffe.

False positive or not applicable? A repository writer can comment:
/ai-review override fable 3e518c04055b924ff847269255b98e88cfe04ffe: <one-sentence reason>

@github-actions

github-actions Bot commented Sep 11, 2026 •

Copy link
Copy Markdown
Contributor

GPT 5.6 Review — ✅ no blocking findings

GPT 5.6 completed its review of 3e518c04055b924ff847269255b98e88cfe04ffe and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] 3e518c0

False positive or not applicable? A repository writer can comment:
/ai-review override gpt 3e518c04055b924ff847269255b98e88cfe04ffe: <one-sentence reason>

@github-actions github-actions Bot added readiness: action required A blocking check or review needs attention and removed readiness: checking Automated validation is still running labels Sep 11, 2026
@Zedmor
Zedmor force-pushed the feat/member-task-board branch from d057b62 to dbe402b Compare September 11, 2026 21:11
@Zedmor

Zedmor commented Sep 11, 2026

Copy link
Copy Markdown
Contributor Author
  • Preserve artifact entries after redaction — fixed; span=7764ccf9d636

Redacted mapping labels now receive a unique suffix when they collide, including collisions with a naturally occurring redaction label. The API regression supplies three colliding labels and verifies all three values survive without exposing credentials.
Implemented in dbe402b.

@Zedmor

Zedmor commented Sep 11, 2026

Copy link
Copy Markdown
Contributor Author
  • Retire closed items without losing history — fixed; span=d27f3d5e5d4a

Owners can archive accepted, rejected, or abandoned tasks, freeing the 32-item active-board capacity. Records and events remain readable through bounded archive pages; bindings and IDs stay reserved, and late reports cannot reopen terminal work. Storage tests cover a full board, retries, conflicts, concurrent archives, and preserved bytes.
Implemented in dbe402b.

@Zedmor

Zedmor commented Sep 11, 2026

Copy link
Copy Markdown
Contributor Author
  • Offer explicit acceptance actions — fixed; span=95f31367e27f

A selected Review task now offers Accept result and Reject result. Each sends the owner decision and exact task context through normal member steering; the UI waits for the ledger verdict and close state instead of moving the card optimistically.
Implemented in dbe402b.

@Zedmor

Zedmor commented Sep 11, 2026

Copy link
Copy Markdown
Contributor Author
  • Replace the private-memory disclaimer — fixed; span=86ea0dda20f4

The introduction now describes creating tasks, giving instructions, and following progress. The unsupported-dispatch premise was also fixed: scoped spawn now binds a private member worker to the ledger before its first turn.
Implemented in dbe402b.

@Zedmor

Zedmor commented Sep 11, 2026

Copy link
Copy Markdown
Contributor Author
  • Use consistent member-facing terminology — fixed; span=98ec4c0ffaab

The interface uses Work session unavailable and Open task conversation. Conductor terminology is confined to the implementation; task steering explicitly names the selected task and targets its member.
Implemented in dbe402b.

@Zedmor

Zedmor commented Sep 11, 2026

Copy link
Copy Markdown
Contributor Author
  • Keep disclosure separate from instruction targeting — fixed; span=fcfa7aaaa6c9

Expanded card state is separate from the retained instruction target. Give instructions explicitly selects the task, a visible label names it, and Clear task selection removes the target. Tests preserve the first draft while another card expands.
Implemented in dbe402b.

@Zedmor

Zedmor commented Sep 11, 2026

Copy link
Copy Markdown
Contributor Author
  • Show each progress summary once — fixed; span=641c11004db3

Collapsed cards show a summary preview; expanded cards show the report once beneath a Progress report label, separate from acceptance criteria. A regression asserts the summary occurs once.
Implemented in dbe402b.

@Zedmor

Zedmor commented Sep 11, 2026

Copy link
Copy Markdown
Contributor Author
  • Match the disclosure chevron to expansion — fixed; span=ee8b23f7793c

Task cards show ChevronRight while collapsed and ChevronDown while expanded, with aria-expanded driven by the same disclosure state.
Implemented in dbe402b.

@Zedmor

Zedmor commented Sep 11, 2026

Copy link
Copy Markdown
Contributor Author
  • Make the progress lanes operational — fixed; span=c24af1e7f2bd

The original user explicitly requested creation, steering, and progress monitoring. Removing the progress lanes would drop that requirement. The implemented repair connects spawn_run(work_item_id=...) to ordinary scoped admission and binds a fresh worker before execution. The integration test exercises progress, blocked, done/open, and accepted states through real HTTP routes.
Implemented in dbe402b.

@Zedmor

Zedmor commented Sep 11, 2026

Copy link
Copy Markdown
Contributor Author
  • Give polling a live report source — fixed; span=5395474a32fc

Private spawned workers now resolve their durable binding through work_brief and work_report after protected process identity verification. The member board sees those real reports; its five-second query remains limited to the visible Tasks tab.
Implemented in dbe402b.

@Zedmor

Zedmor commented Sep 11, 2026

Copy link
Copy Markdown
Contributor Author
  • Retain the work-tool mount for supported member sessions — rebutted; span=567ae56491e9

The default member chat backend is KAS, independently of the global task backend: config/sections.py and config/loader.py default member_acp_backend to kas, and members.select_provider_backend routes member sessions through that field. The existing capability set intentionally covers KAS and Claude; the mount is reachable on the default member path. Scoped workers use the existing worker template. No harness capability set was broadened.

@Zedmor

Zedmor commented Sep 11, 2026

Copy link
Copy Markdown
Contributor Author
  • Use the main composer for general instructions — fixed; span=d18db538ba04

The no-selection instruction form was removed. Task instructions appear only after an explicit task selection; general member messages use the existing conversation composer.
Implemented in dbe402b.

@Zedmor

Zedmor commented Sep 11, 2026

Copy link
Copy Markdown
Contributor Author
  • Remove redundant host-body retention — fixed; span=39c29ff32c91

The keepMounted host-tab option and hidden retained containers were removed. The dashboard draft store retains task drafts and delayed outcomes, so only the active host body needs to mount.
Implemented in dbe402b.

@Zedmor

Zedmor commented Sep 11, 2026

Copy link
Copy Markdown
Contributor Author
  • Use one ordered host-tab interface — fixed; span=92c973dd67d7

SidePanel now accepts only leadingTabs. usePanelTabs accepts only leadingIds. The sole production caller and all affected test harnesses use the ordered interface; the singular compatibility paths were removed.
Implemented in dbe402b.

@Zedmor

Zedmor commented Sep 11, 2026

Copy link
Copy Markdown
Contributor Author
  • Bind and report work for private members — fixed; span=e07a63223a28

The prior intent note incorrectly excluded a mechanism needed by the original request to monitor progress. The repair adds a single-task work_item_id option to scoped spawn. The gateway validates the verified parent, preserves private memory and normal approval, and binds its manager-generated subagent identity before provider execution. The integration test uses real protected files, admission, binding, brief/report routes, and owner acceptance; only the LLM stream and authenticated transport envelope are replaced. Duplicate dispatch, a mismatched caller, and closure during approval are refused.
Implemented in dbe402b.

@chenmingwei23
chenmingwei23 force-pushed the feat/crew-webview branch 12 times, most recently from 558401f to 72ef806 Compare September 19, 2026 20:58
@bolichen97

Copy link
Copy Markdown
Contributor

Maintainer note: PR #11780 (merged 2026-09-19, bc0d2b0cf, "simplify member memory to one SQLite store") removed the protected-binding / private-execution member memory model. Two PRs written against the old model merged after it with a textually clean merge and broke main at import time (#11947, #11960); both are now reverted. This PR carries the same hazard.

Already conflicting; when rebasing please also update test_member_task_dispatch.py:63,131,351, which monkeypatch member_memory_auth._publish_private_binding_dir; that helper is gone.

@chenmingwei23
chenmingwei23 force-pushed the feat/crew-webview branch 8 times, most recently from 4db3d57 to 8af150d Compare September 20, 2026 02:51
@dwu96

dwu96 commented Sep 24, 2026

Copy link
Copy Markdown
Contributor

🤖 Kiro Crew [operator: dwu96#8c08bee0]: This PR has been inactive for 7+ days. I reviewed the blockers but they require your input:

  • The memory/identity model this PR builds on is gone from main. PR #11780 removed the protected-binding / private-execution member memory model. src/kiro_crew/member_memory_auth.py still exists and still exports bind_private_session_store, but _publish_private_binding_dir no longer does. How the PR's stated guarantees — "briefs and reports require the binding and item to name the same worker" and "an interrupted bind cannot authorize an obsolete worker" — should be expressed under the single-SQLite-store model is a design decision, not a conflict resolution. I will not pick that direction for you.
  • Three test sites monkeypatch the removed helper. test_member_task_dispatch.py sets member_memory_auth._publish_private_binding_dir at the three sites @bolichen97 named. Re-pointing them is not a rename: what they assert (private binding-dir publication) may no longer be a behavior that exists, so the replacement depends on your answer above.
  • Confirm the scope still holds after the rebase. This is +9129/-269 across 95 files introducing a new subsystem, and @bolichen97's note records that two PRs written against the old model merged textually clean and broke main at import time (#11947, #11960), both since reverted. A mechanical conflict resolution here reproduces that failure class, so the rebase needs your intent rather than mine.

For the record, nothing else is blocking: the security scan is clean, and all seven review lanes (GPT 5.6, Opus 4.8, First Principles, Design, UX, Automated Rule Check, SAST) pass. PR Readiness is pending only because CI, Fast Gate and Build cannot start while the branch conflicts.

When you've addressed these, the pipeline will re-assess on its next cycle. Add pr-no-autofix to opt out of future automation.

Copy link
Copy Markdown
Contributor

Triage note (automated PR-overlap scan, 2026-10-05). Checked against origin/main @ a97659dfc3.

Status: rebase needed, and keep developing. This is not a duplicate, and main does not already cover it.

  • The branch still carries the pre-squash feat(members): give every crew a configurable webview in its drawer #8468 commit. 0d2ceae522 ("give every crew a configurable webview in its drawer") landed on main as squash d246b0b22a, so agent_panel.py, mcp_panel.py, dashboard/handlers/agent_panel.py and CrewWebview.tsx already exist on main. Only 3e518c0405 (+2406/-187, 55 files) is this PR's own change. A trial merge of the head into main conflicts in 61 files, and the branch is 3063 commits behind. Rebasing just 3e518c0405 onto main should remove most of the conflicts.
  • The feature itself is still missing from main. Main has no agent_panel_templates/tasks.html, spawn_run has no work_item_id, and there is no member_session_servers / MEMBER_SESSION_SERVERS. Nothing merged covers it, fully or in part.
  • The memory-model drift from refactor(memory): simplify member memory to one SQLite store #11780 that @bolichen97 and @dwu96 raised still needs a design answer, not a mechanical conflict fix. The member_memory_auth._publish_private_binding_dir monkeypatch in test_member_task_dispatch.py targets a helper that no longer exists on main.

Overlap to discuss before both land: #16603 ("a project report beside every crewmate's chat") also renders the work ledger in the crewmate Dashboard tab. It adds a built-in work-kanban / project-report template backed by a new workstreams fold. This PR shows a published tasks snapshot that panel_publish reads from the caller's ledger. The goals overlap but the mechanisms differ: a gateway-read published snapshot here, versus a fold-backed manifest template there. Both also modify dashboard/handlers/agent_panel.py and mcp_panel.py. Please agree with the #16603 author on one ledger board for the Dashboard tab, or on how the two templates coexist. #10595 (same author) also shares mcp_work.py, context.py and the members wiring.

This scan found no breaking change and no P0 in this PR's own commit.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

merge conflict Branch has merge conflicts with its base — author must resolve before merge needs-author-decision PR blocked on author input

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants