Skip to content

fix(security): preserve trusted SharePoint links - #13206

Open
rubencu wants to merge 1 commit into
kirodotdev:mainfrom
rubencu:fix/sharepoint-link-redaction
Open

rubencu wants to merge 1 commit into
kirodotdev:mainfrom
rubencu:fix/sharepoint-link-redaction

Conversation

@rubencu

@rubencu rubencu commented Sep 23, 2026 •

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: Preserve provider-generated SharePoint item locators without allowing forged URL text to suppress credential redaction.

A benign 46-character SharePoint item ID can look exactly like a bare AWS secret and be rewritten as [REDACTED: credential]. URL shape and a trusted tenant name alone cannot prove that the bytes came from SharePoint.

Why it matters

The false positive destroys links in rendered messages and persisted history. A text-only carve-out would create a credential-egress path because a model can forge the same URL shape.

No production companion currently supplies the provider-field declaration. The seam remains empty until maintainers decide that companion evidence justifies enabling a specific (server, tool, field) mapping.

Not a goal

  • Trust arbitrary SharePoint URLs, tenants, tool output, or model-authored text.
  • Relax fixed, encoded, token-parameter, query, fragment, or neighboring-secret checks.
  • Invent a production provider-field supplier without maintainer approval and companion evidence.

What changed (motivation → approach → change)

A gateway-only admission path accepts locators only from companion-declared result fields after successful tool completion, resolved MCP identity, retained complete same-call input, exact trusted-host membership, strict SharePoint route validation, and input-echo rejection. Every shared-runtime ACP harness records an explicit identity-readability verdict. Missing evidence, input, or policy fails closed before a worker is scheduled. The public empty policy schedules no admission work.

Accepted locators are stored as canonical SHA-256 digests in a bounded, owner-only, atomic registry under the existing sandbox-hidden gateway authority root. Dashboard and API-only gateways schedule registry preload on a worker thread after readiness without delaying the listener. Cold readers return an empty snapshot, and other process readers use a bounded background file-identity and generation check to adopt cross-process writes. Event-loop snapshot lookups never resolve the registry path: they compare the in-memory home inputs captured with the snapshot, and a changed data-home override returns an empty snapshot until the background load resolves the new path.

The credential redactor stays free of locator-registry imports and I/O. Gateway, transcript, display, and dashboard tool-result boundaries inject immutable digest snapshots; redact() hands its snapshot to every credential pass of the composed pass. The empty default keeps stdio and public callers under markerless credential redaction. Gateway result parsing awaits attestation before redaction.

Scope Purpose
docs/system-specs/modules/acp-client.md, platform-context.md, and security.md Specify provenance admission, process coverage, cross-process refresh, and fail-closed defaults.
src/kiro_crew/acp/locator_attestation.py, the _dispatch.py facade with its _dispatch_parts/ owners, and the session_handle_* owners Record identity readability for every shared-runtime harness and check provenance before offloading persistence.
src/kiro_crew/platform/interfaces.py, defaults.py, and context.py Add the optional exact (server, tool) result-field declaration with an empty public default.
src/kiro_crew/locator_attestations.py and src/kiro_crew/_durable_vouch.py Persist one bounded global digest LRU and publish refreshable immutable snapshots.
src/kiro_crew/dashboard/server.py with its server_runtime/api_entrypoint.py and server_runtime/maintenance.py owners, the chat_utils.py facade with its chat_utils_display.py owner, display_battery.py, and src/kiro_crew/history_append.py Preload both gateway modes and preserve attested links across display, tool-result, and transcript boundaries.
src/kiro_crew/security/redaction.py with its redaction_entropy.py owner, text_redaction.py, and pass_compose.py Validate injected SharePoint digests without registry I/O, hand them to every composed credential pass, and keep the pure document-link helper source-compatible.
src/kiro_crew/security/paths_lists.py Document that the nested registry reuses the existing masked and precreated vouched-executions root; this PR adds no new sealing mechanism.
test/test_sharepoint_locator_attestations.py, ACP launch/composition tests, test_session_handle_bindings.py, test_chat_utils.py, test_display_verdicts.py, and test_tool_call_classification_kas.py Cover no-hop defaults, all-harness identity evidence, off-loop preload and refresh, restart, display caches, and API compatibility.

Backwards compatibility

Compatible: the new adapter declaration is optional and fails to an empty mapping. Public and non-gateway behavior remains full markerless redaction until a companion declares exact provider-generated fields. The one-argument document-link span helper and other signatures visible on main remain callable as before.

Tests

  • Changed attestation, credential-redaction, startup-smoke, display, tool-classification, handle-inventory, server-composition, and security-budget test files: 660 passed and 16 skipped. They include test_dashboard_tool_result_field_keeps_the_link_the_gateway_result_kept, test_credential_beside_an_attested_locator_is_redacted_on_every_composed_pass, and test_tool_field_reads_the_snapshot_once_and_only_for_sharepoint_text, which fail on the previous head because the dashboard tool-result redactor gave the composed pass no snapshot. test_unattested_or_cold_locator_stays_redacted_through_tool_field_and_plain_redact pins the fail-closed side.
  • Broad ACP, redaction, display, platform, startup and app-install overlap across 318 files: 15,144 passed and 68 skipped. Seventeen failures reproduce on unchanged main, all in the branch-unchanged ACP projection-stack class TestRuntimeProjectedLaunchDocuments. Without a built website/dist, all 122 tests in the four server composition contract files pass, with locator preload in the expected post-readiness position.
  • Black, isort, whole-tree Flake8, mypy across 3,540 source files, harness parity, file-line cap, subprocess encoding, ACP frame-host-data, agent SDK boundary, sync-I/O, lockdown-before-publish, comment history, brand, docs lint, and git diff --check passed.
  • Diff-scoped Semgrep loaded 22 code rules and ran 15 Python rules across 36 changed files with zero findings.

Manual verification

N/A. Unit tests cover admission, persistence, cold load, cross-process refresh, restart, batch, stream, neighboring-secret, history, and display-cache behavior.

Screenshots / video

Why no screenshot: The change has no rendered UI delta and only changes provider provenance and credential-redaction decisions.

Related Issues

No public issue links this security defect. Maintainer disposition is required before this zero-supplier production seam can be enabled.

Pattern harvest

Rule candidate: review-prompt

Pattern: a credential heuristic can exempt provider bytes only after a non-agent-writable boundary attests exact provenance. URL text and host membership are claims, not proof.

Checklist

  • At most two commits, with a Conventional Commits title
  • Existing tests pass and new tests cover the changed behavior
  • Self-review completed and code follows project style guidelines
  • Security, platform, and ACP documentation updated
  • No secrets, credentials, or internal references in the diff

Contribution License Agreement

Contributions are accepted under the repository's Apache 2.0 license.

@rubencu
rubencu requested review from a team and buluoray September 23, 2026 23:12
@rubencu

rubencu commented Sep 23, 2026

Copy link
Copy Markdown
Contributor Author

Intent: Preserve generated SharePoint item locators on policy-owned exact trusted tenants without weakening higher-confidence credential redaction or markerless detection elsewhere.
Not a goal: Exempt arbitrary document hosts, arbitrary SharePoint tenants, or general URL path tokens from credential scanning.

@github-actions github-actions Bot added readiness: action required A blocking check or review needs attention fork Pull request from a fork (external contributor) labels Sep 23, 2026
@github-actions

github-actions Bot commented Sep 23, 2026 •

Copy link
Copy Markdown
Contributor

First Principles Review (Opus 5.5, fork) — 🔴 BLOCK

Premise-level review of 5d9c69e6a52af89784fd4d6fbfed09aec4ac6c2a via the fork AI-review pipeline — 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 does not.

First-Principles-Verdict: BLOCK

The whole attestation mechanism ships turned off. The only supplier returns {}, and the description says "No production companion currently supplies the provider-field declaration." So no user's SharePoint link survives redaction any more than it does today.

Notes

  • Item 6 — rides along: every harness's tool-call frames now record an identity-readability verdict, where before only KAS frames did. That changes how every non-KAS harness handles tool calls. The only reader is the dormant admission path.

What this change ships

Inventory (8 items) — 0 justified

Intent: keep provider-generated SharePoint links readable instead of showing them as [REDACTED: credential]. This is a FIX.

  1. Benign SharePoint links no longer redacted — zero consumers (no supplier declares fields)
  2. New CredentialPolicy.provider_generated_sharepoint_fields() companion method — zero consumers
  3. Persisted digest registry under vouched-executions/locator-attestations/ — zero consumers
  4. Every dashboard/API gateway boot schedules a registry preload thread — zero consumers
  5. New attested_sharepoint_digests parameter on redact, redact_credentials, display_battery — zero consumers
  6. Every harness records an identity-readability verdict per tool call — rides along
  7. Gateway awaits an attestation check before each tool-result redaction — zero consumers
  8. Display-cache key gains a locator generation — zero consumers

Blockers

Dormant seam: the fix ships switched off — zero consumers
Trigger: DefaultCredentialPolicy.provider_generated_sharepoint_fields returns {}. In src/, def exempt_exact_hosts has 2 sites (the Protocol and the default), so this repo has no other policy that could supply fields. The description says enabling one "remains an explicit maintainer decision".
Outcome: about 2,800 lines, a new public companion method, a persisted file format and a boot-time task have to be maintained forever, and the reported false positive is not fixed for anyone.
Subtraction: drop items 2–8 until a real supplier ships in the same change.
Clears when: the PR includes a non-empty provider-field supplier that a production path uses, or the dormant mechanism is removed.

[FIRST-PRINCIPLES-REVIEWED] 5d9c69e

Description read

[DESCRIPTION-READ] 20f9792136e2ac08badaf6c0030429d751a7d8328c8177c11cf2a4d7b6408cf4

That sha256 names the PR description this verdict read. Recompute it with .github/scripts/pr-description-capture.sh; if it differs, the description changed after this verdict was formed and any finding drawn from the description is unproven.

@github-actions

github-actions Bot commented Sep 23, 2026 •

Copy link
Copy Markdown
Contributor

Design Review (Opus 5.5, fork) — ✅ PASS

Design-level review of 5d9c69e6a52af89784fd4d6fbfed09aec4ac6c2a via the fork AI-review pipeline — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS does not.

Design-Verdict: PASS

Nothing blocks: the empty public policy keeps every path failing closed, and the identity-cache split keeps KAS-only refusal semantics intact.

Notes

  • This ships about 3,000 lines of dormant machinery with no supplier: a durable registry, a boot preload task and a per-second background revalidation thread. The PR itself says "No production companion currently supplies the provider-field declaration", and the public exempt_exact_hosts() is empty, so nothing in this tree can ever attest. Getting maintainer disposition before merge, not after, would avoid carrying unexercised authority code.
  • _identity_unreadable_cache_for_wire now records verdicts for every harness, and the KAS signal moves to _identity_unreadable_cache_for_permission. Any future caller that reads identity_unreadable_cache is not None as "KAS" (permission_requests.py:1374) has to use the permission view. A pinning test for that seam would stop it from silently widening KAS-id folding to other harnesses.

[DESIGN-REVIEWED] 5d9c69e

Description read

[DESCRIPTION-READ] 3e442264b4138dae1b510a511e6faefaec816609c5b44e691a21c39e7b2e2a9c

That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

github-actions Bot commented Sep 23, 2026 •

Copy link
Copy Markdown
Contributor

GPT 6.1 Review (fork) — 🔴 changes requested (blocking)

Reviewed 5d9c69e6a52af89784fd4d6fbfed09aec4ac6c2a via the fork AI-review pipeline; updated in place on each push.

1 of 1 blocking finding(s) are security-class and were withheld from adjudication, so the blocking verdict stands.

BLOCKING -- test/test_chat_utils.py:522 -- Drain the locator loader before home isolation unwinds

generation = str(chat_utils.locator_attestations.snapshot()[0]).encode("ascii")
Trigger: A cold snapshot starts locator-attestation-load; when that worker reaches _refresh_from_disk() after this test’s home pin unwinds, _path() → data_home() → config_dir() resolves the subsequent test’s home. Shared teardown does not join this daemon, and _isolated_registry applies only to the new attestation test module.
Outcome: config_dir() creates directories outside the originating test’s isolation, introducing filesystem mutations into another test.
Anchor: no-test-side-effects
Fix: Join locator-load workers in the shared pre-fixture teardown, before home pins lift, alongside the existing consent-worker drain.

[BLOCK-MERGE] 5d9c69e
[GPT-REVIEWED] 5d9c69e

Adjudication (Opus 5.5) — is blocking on each finding proportionate?

The adjudicable block has no findings, and the one fenced finding (F1) stays blocking as UPHOLD-FENCED. I did not FLAG it: a FLAG needs the conditions to be rare in practice, and the repository has already recorded this exact pattern causing a real CI flake.

F1 — UPHOLD-FENCED. Each step it needs is real, in the trusted base or the patch:

  • The thread: the patch's snapshot() (authentic.patch:1057) starts a daemon thread named locator-attestation-load (authentic.patch:1049-1052). That thread calls _refresh_from_disk (authentic.patch:1001-1007), which calls _path() and data_home() when the thread runs, not when it was scheduled.

  • The directory creation: with a valid override, data_home() calls config_dir() (src/kiro_crew/config/paths.py:823-824), and config_dir() creates whatever home is pinned at that moment (src/kiro_crew/config/paths.py:768).

  • No drain: conftest.py:736-785 only joins the telemetry consent worker, so nothing joins this new thread before the home pins lift.

  • Precedent: the docstring at conftest.py:745-759 describes the same late-running worker creating the next test's home, measured on a Windows shard, with a named flaky test as the symptom. So the conditions are common, not extreme.

  • The fix is small: join one more thread, by name, in the existing pre-teardown drain.

    [ADJUDICATION] 5d9c69e total=0 uphold=0 downgrade=0
    [GPT-ADJUDICATED] 5d9c69e

    [ADJUDICATION-FENCED] 5d9c69e fenced=1 flagged=0
    UPHOLD-FENCED F1 test/test_chat_utils.py:522 -- The undrained locator-attestation-load daemon resolves config_dir() when it runs, after the scheduling test's pins lift, the same shape conftest.py:745-759 records as a measured CI home-creation flake, so the conditions are plausible rather than extreme.
    [GPT-ADJUDICATED-FENCED] 5d9c69e

@github-actions

github-actions Bot commented Sep 24, 2026 •

Copy link
Copy Markdown
Contributor

Opus 5.5 Review (fork) — ✅ no blocking findings

Reviewed 5d9c69e6a52af89784fd4d6fbfed09aec4ac6c2a via the fork AI-review pipeline; updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] 5d9c69e

@github-actions

github-actions Bot commented Sep 24, 2026 •

Copy link
Copy Markdown
Contributor

Security Scope Review (fork) — ✅ PASS

Which legitimate operations does 5d9c69e6a52af89784fd4d6fbfed09aec4ac6c2a newly refuse? Model proposes, scripts/deny_diff.py decides — updated in place on each push. A script-confirmed regression blocks PR readiness.

adjudicated, zero confirmed regressions

Scope-Verdict: PASS

Reason: The change does not tighten the shell deny composite. It touches no sensitive-bash, exfil or deny-rules code. The only edit to a deny-path list is the comment on "vouched-executions" in _CREW_SECRET_LEAVES. The one guard that does get stricter is the KAS-agnostic identity-readability requirement inside gateway locator attestation. It sits outside the classifier, so it is named under UNADJUDICATED rather than passed on an empty adjudicated set.

TIGHTENING: gateway-side locator-attestation admission (and the readability cache it depends on) -- it refuses to treat a SharePoint link that an agent wrote, or echoed from the call input, as a provider-generated locator: or cache_key not in identity_unreadable_cache / or identity_unreadable_cache[cache_key] is not False in attest_session_update, plus _SHAREPOINT_LOCATOR_RE and the same-call input scan in locator_attestations.py. On the shell side the only hunk is - "vouched-executions", # member-store vouches; also sandbox-masked / + "vouched-executions", # gateway-minted vouches and locator attestations, a comment-only edit. The new registry vouched-executions/locator-attestations/sharepoint.json sits under a leaf the base already refuses.

ADJUDICATED: 13 candidates, 0 confirmed newly-refused

PLATFORM COVERAGE: posix, windows and any. Posix and windows rows cover the path and executable spellings near the touched area: the new and touched test modules through .venv/bin/python and .venv\Scripts\python.exe, a SharePoint link HEAD check through curl and curl.exe with $ITEM_ID and %ITEM_ID%, a spec search through grep and findstr, and a listing of the sandbox-writable trust/ sibling under $KIROCREW_HOME and %KIROCREW_HOME%. One any row (kirocrew restart) covers the gateway restart that now schedules the preload. The diff's own platform assumptions are in file I/O: an IS_WINDOWS stat/read branch, and O_NOFOLLOW/fstat on posix. A shell row cannot reach those, so they appear only under UNADJUDICATED.

UNADJUDICATED: Three guards are outside the shell classifier:

  1. Gateway attestation admission.

    • Call path: session_handle_dispatch._dispatch_events (an "update" action, or a subagent update) → attest_session_update → sharepoint_attestation_policy(server, tool) → asyncio.to_thread(attest_tool_result) → locator_attestations.attest_provider_result.
    • Inputs I would have submitted:
      • A completed tool_call_update from a non-KAS shared-runtime harness whose tool_call frame arrived before the update. This checks that the explicit False is now written for every harness, so a legitimate locator is not refused for lack of a verdict.
      • The same call when a child session's tool_call frame and its update use different sessionId scopes.
      • A provider result whose declared field holds an uppercase-host link, https://Contoso.SharePoint.com/:x:/s/Team/E…. The regex is IGNORECASE and the digest lowercases the host, but the trusted_hosts comparison should be checked against an upper-case policy entry.
      • A site segment carrying %20, or one longer than 128 characters (the {1,128} bound).

    The public default provider_generated_sharepoint_fields() returns {}, so no locator is admitted in the public edition either before or after this change.

  2. KAS permission refusal view. The path is _build_permission_event → _identity_unreadable_cache_for_permission(), which still gates on ACP_BACKENDS_PERMISSION_KIND_FROM_TOOL_CALL. The input to submit is a non-KAS harness permission request after a tool_call frame whose identity was readable. Those frames now write an explicit False instead of popping the entry, and the check is that this does not turn into a refusal.

  3. Registry file reads on each platform. The path is _read_file → refuse_linked_parent → either the Windows os.stat/read_bytes_with_retry branch or the posix O_NOFOLLOW branch. Inputs to submit:

    • A data home under a Windows UNC path or a drive-letter path.
    • A posix data home reached through a symlinked home directory.

    A refusal at either point only leaves ordinary redaction in place. That is the fail-closed direction, not a lost operation.

[SCOPE-REVIEWED] 5d9c69e

Security scope review -- adjudicated candidates

  • macos-latest (platform posix): 9 classified, 0 newly refused, 8 still allowed.
  • ubuntu-latest (platform posix): 9 classified, 0 newly refused, 8 still allowed.
  • windows-latest (platform windows): 5 classified, 0 newly refused, 4 still allowed.

No candidate flipped from allowed to refused.

@github-actions github-actions Bot added the merge conflict Branch has merge conflicts with its base — author must resolve before merge label Sep 25, 2026
@cruisercohen cruisercohen added the needs-pr-triage PR scanner: awaiting automated triage label Sep 30, 2026
@NicholasRBowers NicholasRBowers added needs-author-decision PR blocked on author input and removed needs-pr-triage PR scanner: awaiting automated triage labels Sep 30, 2026
@NicholasRBowers

Copy link
Copy Markdown
Contributor

🤖 Kiro Crew [operator: NicholasRBowers#a942f9ca]: This PR has been inactive for 7+ days. I reviewed the blockers but they require your input:

  • The GPT 5.6 BLOCK objects to the premise, not the implementation, so no automated fix can clear it. It holds that an AWS secret wrapped in a forged trusted-tenant SharePoint locator skips pass 3 and leaks (redaction.py:1367), and the Opus adjudicator upheld it as unbounded harm, noting the PR's own test fixture builds the attack. Its remedy — "require trusted out-of-band connector provenance; do not exempt based solely on URL text" — replaces the whole mechanism. The only other route is a maintainer override accepting the residual you already document, and that is a policy decision about which link shapes are benign.
  • Design Review's Watch item needs a fact only you have: on public installs exempt_exact_hosts() returns an empty set, so the fix does nothing there. Please confirm the reporting environment has a companion adapter, or file a follow-up for a trust source standalone installs can use.
  • A maintainer has to pick the host PR. fix(security): stop redacting shared document links as bare secrets #9811 addresses the same problem (issue [bug] KiroCrew Security Module redacts legit Google Workspace links because it looks like credentials in the URL. #11364), is blocked by the same class of security finding, and was triaged needs-author for the same "which link shapes are benign" question. Whichever PR carries the fix should get the Fixes #11364 trailer.
  • Opus's two advisory findings both widen the exemption, which pulls against the GPT block, so they need deciding together rather than one at a time: the whole-run containment check only preserves ids whose -/_ falls in the first ~6 characters, so most real links are still redacted; and the ?e= lookahead rejects trailing punctuation and Slack <url|text> links.

The remaining items are mechanical and can wait for that ruling: re-porting the pass-3 decline into #12684's _credential_redaction_plan, and syncing the four exempt-host scope docstrings Opus names. The Backend Lint red is not yours either — the job hit its 25-minute ceiling with Black taking ~16 minutes, and every step including isort, flake8 and mypy reported success.

When you've addressed these, the pipeline will re-assess on its next cycle.

Copy link
Copy Markdown
Contributor

PR triage (open PRs against main): needs a rebase, and the overlap needs discussing

I checked the head 85e06a1 against origin/main 5edf2fe. The PR is 1,861 commits behind its merge base fddab6f, and git merge-tree reports conflicts in src/kiro_crew/security/redaction.py and docs/system-specs/modules/security.md.

What changed on main since the last triage note

Functional gap I reproduced on this head (advisory, not P0). I trusted contoso.sharepoint.com and used a random 46-char I… id with a single _. The link is preserved when the _ sits at index 5-30. It is still redacted when the _ sits at index ~40, because the run /s/LaunchMaterials/I<40 chars> starts before share_start, so the whole-run containment check at redaction.py:1367 fails. This is the same area as Opus's first finding.

Not raising a new blocker. The open GPT 5.6 BLOCK (forged trusted-tenant locator skips pass 3) already covers the security concern. There is no breaking change: the default exempt_exact_hosts() is still frozenset() and the Protocol signature is unchanged.

Recommendation: REBASE onto main, then merge discussion with the owners of #15792 / #16952 about one shared pass-3 link-decline mechanism. After that, continue development on the late-separator gap.

@rubencu
rubencu force-pushed the fix/sharepoint-link-redaction branch from 85e06a1 to 7b1d3ad Compare October 14, 2026 21:39
@rubencu
rubencu requested review from a team and KiroCrewCaptain as code owners October 14, 2026 21:39
@rubencu

rubencu commented Oct 14, 2026

Copy link
Copy Markdown
Contributor Author

Rebuilt on current main at 37449b3 with the accepted durable-attestation design. URL text alone grants no exemption: only gateway-resolved, companion-declared provider fields can persist bounded SHA-256 locator digests in the sealed leaf, and redaction.py no longer imports exfil or platform. Parent verification passed 123 attestation/display/flake tests; the broader focused run passed 533. Body accounting, push guard, and the fresh lease passed on 7b1d3ad. CI and security review are pending.

@github-actions github-actions Bot added readiness: checking Automated validation is still running and removed merge conflict Branch has merge conflicts with its base — author must resolve before merge readiness: action required A blocking check or review needs attention labels Oct 14, 2026
@github-actions

github-actions Bot commented Oct 14, 2026 •

Copy link
Copy Markdown
Contributor

UX Review (Opus 5.5, fork) — ⏭️ skipped

Revision 5d9c69e6a52af89784fd4d6fbfed09aec4ac6c2a touches no user-facing surface (no changes under website/ or committed screenshots), so the UX review was skipped. Does not block merge.

@github-actions github-actions Bot added readiness: action required A blocking check or review needs attention merge conflict Branch has merge conflicts with its base — author must resolve before merge and removed readiness: checking Automated validation is still running labels Oct 14, 2026
@rubencu
rubencu force-pushed the fix/sharepoint-link-redaction branch from 7b1d3ad to a2f03f4 Compare October 15, 2026 01:00
@rubencu

rubencu commented Oct 15, 2026

Copy link
Copy Markdown
Contributor Author
  • span=1327b31bcbc3 fixed by subtraction

redaction_policy.py was removed and current main’s sanctioned function-local token-prefix policy read is restored.

@rubencu

rubencu commented Oct 23, 2026

Copy link
Copy Markdown
Contributor Author

Pushed 2941b0b, rebased onto 7796e2a.

  • GPT's blocking finding on 040618a (surfaced_text.py:213): _redact_tool_field called security.redact with no attestation snapshot, so an attested SharePoint locator in a completed declared-field tool result became a credential marker before the broadcast and before it was stored in message metadata. _redact_tool_field now reads the snapshot once with locator_attestations.digests_for_text, which reads nothing for text that names no SharePoint host. It passes the snapshot through redact and redact_with_findings into all three credential passes of pass_compose.compose. Every other redact() caller keeps the empty default and stays fail closed. New tests in test_sharepoint_locator_attestations.py:
    • test_dashboard_tool_result_field_keeps_the_link_the_gateway_result_kept
    • test_credential_beside_an_attested_locator_is_redacted_on_every_composed_pass
    • test_tool_field_reads_the_snapshot_once_and_only_for_sharepoint_text
    • test_unattested_or_cold_locator_stays_redacted_through_tool_field_and_plain_redact, a guard that passes on both heads
      The first three fail on 040618a.
  • Still fail closed, and not named by the finding: the native sub-agent card and the PostToolUse hook redact their copy of the tool output with plain redact_credentials, so an attested link is still masked there.
  • The body now says the empty public policy "schedules no admission work", answering the Design note that every render takes a snapshot. Skipping that snapshot when no provider fields are declared depends on whether the pipeline stays, which is the open First Principles question for maintainers.
  • Conflict with main's fix(security): deny kirocrew skills authority-retire from an agent shell #4533 in test_security_gate_liveness.py: both notes are kept, and the budget was re-measured at 32,266.
  • Scoped tests on 939e0e3: 15,144 passed, 68 skipped; the 17 TestRuntimeProjectedLaunchDocuments failures also fail on a clean 939e0e3 checkout. The PR's changed test files: 660 passed, 16 skipped. The dist-less composition contract (65), the static gates and the 2,000-line cap passed. On 1f03360: 3,351 PR and main-touched tests passed. On the final base: 595 PR tests and the cap passed again.

@rubencu

rubencu commented Oct 23, 2026

Copy link
Copy Markdown
Contributor Author

Pushed e06cdd9, rebased onto 99cced3.

  • Diff-owned red on 20f14cb, Backend shard 2 and Windows shard 2: main's feat: add Coder remote session environments #19833 test test_remote_session_local_only_tools.py::test_the_session_handle_names_the_workspace_only_when_its_runtime_attached_it builds its session handle as a SimpleNamespace that has only _identity_unreadable_cache_for_wire. This PR's _build_permission_event (session_handle_permissions.py:605) calls _identity_unreadable_cache_for_permission(), the KAS-only view of the shared readability cache, so the stand-in raised AttributeError. The stand-in now also carries _identity_unreadable_cache_for_permission=lambda: None; the test's assertions are unchanged.
  • The other Windows reds on 20f14cb were main-side. The Paths don't have the same drive failures in test_coder_archive_write_races.py and test_remote_workspace_file_integrity.py are fixed on main by fix(workspace-tools): guard cross-drive commonpath ValueError on Windows #23405, which this rebase carries. The test_proof_line_check.py bash cases are main's, from feat(pr-hygiene): ask fix PRs for a Proof line inside Tests #22971.
  • GPT and Opus had no findings on 20f14cb. The First Principles gate (no supplier yet) is unchanged.
  • On a3e2a46: 873 PR and main-touched tests passed, as did the remote-session, bindings, KAS and attestation files, the dist-less composition contract (65), the static gates and the 2,000-line cap. On the final base: 674 PR tests and the cap passed again.

@rubencu

rubencu commented Oct 23, 2026

Copy link
Copy Markdown
Contributor Author

Pushed 6462323, rebased onto 5f77c3f.

  • Conflict with main's fix(security): block shell-expansion (arithmetic) evasion of the IMDS denial #22826 (shell-expansion IMDS gate) in test/test_security_gate_liveness.py. Main re-measured the security package budget from 32,206 to 32,438. Both notes are kept, and the PR's budget was re-measured at 32,498: main's 32,438 plus this PR's unchanged 60 lines. Every other PR line matches the published patch.
  • On e06cdd9 the remote-session stand-in red is gone. The remaining reds are not this PR's:
    • Backend shard 7: test_config_loader_default_agent.py::TestDanglingDefaultAgentReset raised unraisable-exception warnings. The same test went red on another PR the same way, and this PR touches no config code.
    • Windows shard 1: main's test_proof_line_check.py bash cases.
    • Windows shard 2: test_auto_improvement_members.py timeouts.
    • Also red: the recorded First Principles gate, Sensitive Change Review (fork infra) and Coverage Gate downstream.
  • GPT and Opus had no findings.
  • 1,164 PR, budget, remote-session, main-touched and composition tests, the static gates and the 2,000-line cap passed.

@rubencu

rubencu commented Oct 23, 2026

Copy link
Copy Markdown
Contributor Author

Rebased onto main 33cdb8f1. Main's #23335 raised the security package budget in test/test_security_gate_liveness.py from 32,438 to 32,447 (one paths_lists entry), so the rebase conflicted on _PACKAGE_LINE_BUDGET.

Resolution: both budget notes kept. This PR's note now starts from 32,447, and the budget is re-measured at 32,507 (main's 32,447 plus this PR's 60 lines, unchanged). Main's paths_lists and platform-context.md edits merged beside this PR's without conflict. The other 39 paths in the patch are identical to the previous head.

The rebase also picks up two main fixes for Windows reds on the previous head, neither from this diff: #23449 (the test_session_memory_prefetch.py deadline float flake, Windows shard 4) and #23445 (the test_proof_line_check.py Git Bash template cases, Windows shard 1).

Checks on 1f1ba9b4:

  • The PR's tests, the tests main changed since the previous base, and the platform, posture, locator and security modules (220 files): 17,364 passed. The 11 failures are this host's environment and not from this diff: 10 also fail on a clean main checkout, and the eleventh is the boot-order pin, which needs a built website/dist.
  • Dist-less dashboard composition (65), static gates, the 2,000-line cap, docs lint and the PR body check all pass.

@rubencu

rubencu commented Oct 23, 2026

Copy link
Copy Markdown
Contributor Author

Rebased onto main 44702cbc. Main's #23292 (the rm_floor backtick and glued-redirect reads) raised the security package budget in test/test_security_gate_liveness.py from 32,447 to 32,518, so the rebase conflicted on _PACKAGE_LINE_BUDGET again.

Resolution: both budget notes kept. This PR's note now starts from 32,518, and the budget is re-measured at 32,578 (main's 32,518 plus this PR's 60 lines, unchanged). Main's edits to files this PR also changes merged without conflict: acp/_dispatch.py and acp/_dispatch_parts/permission_requests.py (#23452, #23498), dashboard/server.py and the composition reach test (#23010), and acp-client.md (#19905). The other 39 paths in the patch are identical to the previous head.

Checks on the rebase:

  • The PR's tests, the tests main changed since the previous base, and the ACP dispatch, permission-request, remote-session, dashboard-server, platform, posture and security modules (557 files): 32,576 passed. The 13 failures are not from this diff:
    • 11 are this host's environment (auto-improvement members, design-tweak devproc, managed launcher restart, search segmentation), recorded earlier on clean main checkouts.
    • The boot-order pin fails only with a built website/dist; the dist-less run passes (65).
    • test_crewmate_feed.py (one case) timed out under the parallel run and passed alone (35 of 35, twice).
  • After main moved to 44702cbc: the PR's tests, the newly touched tests and the budget test passed (749, plus the same boot-order pin).
  • Static gates, the 2,000-line cap, docs lint and the PR body check pass.

@rubencu

rubencu commented Oct 24, 2026

Copy link
Copy Markdown
Contributor Author

Rebased onto main 177acfe6. Main's #23146 (plural-route commit permalinks in the bare-secret pass) rewrote the "Bare label-less secret-key detection" paragraph in docs/system-specs/modules/security.md, and this PR adds its "Provider-attested SharePoint locators" bullet directly after that paragraph, so the rebase conflicted there.

Resolution: main's rewritten paragraph is kept, followed by this PR's SharePoint bullet, byte-identical to the previous head's. Main's other edits to files this PR also changes merged without conflict: security/redaction_entropy.py (#23146) and the composition reach test (#22833). The measured security package total is still 32,578, matching this PR's budget pin. Compared with the previous head, all 40 paths in the patch carry the same changes.

Checks on the rebase:

  • The PR's tests, the tests main changed since the previous base, and the ACP dispatch, permission-request, remote-session, dashboard-server, platform, posture and security modules (744 files): 40,814 passed. The 25 failures are this host's environment and not from this diff:
    • 13 in test_cli_manifest_signature.py (an unexpected uv download URL), which also fail on a clean 177acfe6 checkout.
    • 11 more recorded on clean main checkouts earlier: auto-improvement members, design-tweak devproc, managed launcher restart and search segmentation.
    • The boot-order pin, which needs a built website/dist; the dist-less run passes (65).
  • Static gates, the 2,000-line cap, docs lint and the PR body check pass.

@rubencu

rubencu commented Oct 24, 2026

Copy link
Copy Markdown
Contributor Author

Rebased onto 9ae6bc7124 (head 647c105aa0). The only conflict was the security package line budget in test/test_security_gate_liveness.py. Main #23627 raised it to 32,526 for one _WRITE_PROTECTED_HOME_PATHS entry.

  • Budget: both notes are kept, and this PR's note now reads "Raised again, from 32,526". I re-measured the package on the rebased tree at 32,586 lines (main plus this PR's unchanged 60) and set the pin to that.
  • paths_lists.py: main's new entry and this PR's vouched-executions comment merged side by side.
  • Other paths: the remaining 38 carry the same diff as before.

Tests:

  • Scoped run: 43,633 tests passed. The 21 failures are host-state reds that reproduce on a clean main checkout here:
    • 13 in test_cli_manifest_signature
    • 6 in test_managed_launcher_restart
    • test_one_dictionary_per_host
    • the boot-order pin, which needs a built website/dist and passes without one (65/65)
  • After the final rebase: main's newly touched tests and this PR's tests passed, apart from the same pin.
  • Checks: the static gates, line cap, docs lint and check-body pass.

@rubencu

rubencu commented Oct 24, 2026

Copy link
Copy Markdown
Contributor Author

Rebased onto ebe680a698; the new head is 35bfa55920. The rebase had no textual conflict, but one pin needed a change:

  • Lazy-import count pin in test/test_dashboard_server_composition_contract.py. Main feat: ignore local enables with no governance key in enterprise builds #23703 added kiro_crew.edition_locks: local_enable_locked and raised the count from 92 to 93. This PR had made the same 92 to 93 edit for locator_attestations, so git merged the two edits as one. The tree now has 94 lazy imports, and the pin is now 94.

Main's edits to test/test_chat_utils.py, test/test_dashboard_server_composition_contract_reach.py and platform-context.md merged cleanly. The security package budget (32,586 lines) is unchanged.

Tests:

  • On 56b568a2be, a 1,047-file scoped selection passed 50,969. Besides the pin above, the failures were all host state: 18 gallery-dispatch setup errors, 6 managed-launcher restart cases, 10 doctor cases that fail the same way on a clean main checkout on this host, and the boot-order pin.
  • Main moved twice more while the tests ran. On each new base I re-ran the tests main had touched plus this PR's tests: 13,314 passed on 8956047af4, then 1,585 on ebe680a698, plus 655 tool-result and context tests. The only failure each time was the boot-order pin, which needs a built website/dist and passes without one (65/65).
  • Static gates, the line cap, docs lint and the body check passed.

@rubencu

rubencu commented Oct 25, 2026

Copy link
Copy Markdown
Contributor Author

Rebased onto c4ce1efa5d; the new head is 4219fee5df. There was one conflict, the security package line budget in test/test_security_gate_liveness.py:

  • Main fix(security): keep member records in a gateway-only directory #23645 raised the budget by 1, from 32,526 to 32,527, for its member-records leaf in paths_lists. Main's note and this PR's note are both kept, and this PR's note now reads "Raised again, from 32,527". The package now measures 32,587 lines: main's 32,527 plus this PR's unchanged 60. The pin is now 32,587.

Main's edits to security/paths_lists.py and security.md merged cleanly, and the lazy-import pin (94) is unchanged.

Tests, run on 3503760e2e:

  • A 436-file scoped selection passed 26,357 tests. The failures were all host state: 18 gallery-dispatch setup errors, and the boot-order pin, which needs a built website/dist and passes without one (65/65).
  • The static gates, the line cap, docs lint and the body check passed.

Main then moved to c4ce1efa5d (#23815, tests only, no shared file). On that base, the two tests it touched, plus this PR's tests, passed (908, and 65 without a built website/dist).

@rubencu

rubencu commented Oct 25, 2026

Copy link
Copy Markdown
Contributor Author

Rebased onto b784bbc7f9; the new head is 5b1e38a99a. There was one conflict, a count pin:

  • Lazy-import count pin in test/test_dashboard_server_composition_contract.py. Main's own lazy-import count is now 95, and this PR still adds locator_attestations, so the pin is now 96. The composition contracts pass without a built website/dist (65/65).

The rest of the patch merged without conflicts. That includes main's edits to dashboard/server.py, server_runtime/api_entrypoint.py, the reach test and the two docs. The security package budget (32,587 lines) is unchanged.

Tests, run on 289eb79ee3:

  • A 606-file scoped selection passed 35,197. The failures were all host state:
    • 18 gallery-dispatch setup errors.
    • The boot-order pin, which needs a built website/dist and passes without one (65/65).
    • test_governance_policy_viewer.py::TestNoPolicyDefault::test_enterprise_unnamed_scope_reports_the_baseline. This host resolves to the enterprise profile and can't fetch a central policy, and the test fails the same way on a clean main checkout here.
  • The static gates, the line cap, docs lint and the body check passed.

Main then moved twice more: #20821, #22272 and #22300, then #23538, #23550, #23712, #20410, #21222 and #23725. Neither batch touched a file this PR changes. On each base I re-ran the tests those commits touched plus this PR's tests: 749, then 1,037 passed, with the composition contracts at 65 both times.

@rubencu

rubencu commented Oct 25, 2026

Copy link
Copy Markdown
Contributor Author

Rebased onto badd8f5aa8; the new head is c72dadfd8e. There was one conflict, the security package line budget in test/test_security_gate_liveness.py:

  • Main feat(notifications): Web Push for the notification bus #23333 (Web Push) raised the budget by 3, from 32,527 to 32,530, for its web-push leaf in paths_lists. Main's note and this PR's note are both kept, and this PR's note now reads "Raised again, from 32,530". The package now measures 32,590 lines: main's 32,530 plus this PR's unchanged 60. The pin is now 32,590.

The rest of the patch merged without conflicts.

Tests, run on 19877bc051:

  • A 199-file scoped selection passed 9,140 tests. It covers this PR's tests, the tests main touched since the previous base, and the redaction, display and composition overlap. Four failed, all specific to this host: three test_cli.py::TestDoctor cases, which fail the same way without this PR, and the boot-order pin, which needs a built website/dist and passes without one (65/65).
  • The security liveness file (10/10), the static gates, the line cap and the description check passed.

Main then moved to badd8f5aa8 (#24052, testing/harness.py only; no shared file). On that base, test/test_harness.py and this PR's tests passed (838, the same boot-order pin aside, and 65 without a built website/dist).

@rubencu

rubencu commented Oct 26, 2026

Copy link
Copy Markdown
Contributor Author

Rebased onto 9d31193afe; the new head is d319ef53ad. Main touched two of this PR's pins:

  • Security package line budget in test/test_security_gate_liveness.py (conflict). Main perf(redaction): native single-pass credential gate, Python fallback #19628 (native single-pass credential gate) raised it by 71, from 32,530 to 32,601. Main's note and this PR's note are both kept, and this PR's note now reads "Raised again, from 32,601". The package now measures 32,661 lines: main's 32,601 plus this PR's unchanged 60.
  • Lazy-import count in test/test_dashboard_server_composition_contract.py. It merged without a conflict but was wrong after the merge: main fix: exit an enterprise boot that has no kiro login or no governance rules #23968 (kiro_crew.enterprise_boot → require_kiro_login) and this PR (locator_attestations) each moved assert len(expected) == 95 to == 96, so git kept one copy of the identical edit. The table now holds 97 entries, and the pin now says so. With the merged value the test fails (assert 97 == 96).

The other 38 paths have the same diff as on c72dadfd8e, including dashboard/server.py, server_runtime/api_entrypoint.py, platform-context.md and security.md, which main also edited.

Tests:

  • On 46557948 (before main's last move), a 322-file scoped selection passed 20,400 tests: this PR's tests, the redaction, credential and attestation suites (with main's native-gate fallback and pre-filter suites, which run the Python gate here), and the tests main touched since the previous base. Four failed, all specific to this host: three test_cli.py::TestDoctor cases, which fail the same way without this PR, and the boot-order pin, which needs a built website/dist and passes without one. The liveness file, static gates, line cap and description check passed.
  • On 9d31193afe, the 52 test files main touched since 46557948 plus this PR's tests passed 3,249 with two failures: the lazy-import pin (corrected above) and the same boot-order pin. After the correction, this PR's tests pass (751, that pin aside), all 65 pass without a built website/dist, and docs lint and the line cap pass.

@rubencu

rubencu commented Oct 26, 2026

Copy link
Copy Markdown
Contributor Author

Rebased onto 2356a1ba1c; the new head is 2e06f42412. There was one conflict, the security package line budget in test/test_security_gate_liveness.py:

  • Main fix(security): extend native config write tier #23935 (native config write tier) raised the budget by 12, from 32,601 to 32,613, for the crewmate folder's state files in paths_lists. Main's note and this PR's note are both kept, and this PR's note now reads "Raised again, from 32,613". The package now measures 32,673 lines: main's 32,613 plus this PR's unchanged 60. The pin is now 32,673.

The rest of the patch merged without conflicts, including src/kiro_crew/security/paths_lists.py, src/kiro_crew/dashboard/server.py, docs/system-specs/modules/security.md and docs/system-specs/modules/platform-context.md beside main's edits: the other 39 paths have the same diff as on d10f7d1567.

Tests, run on 2356a1ba1c:

@rubencu

rubencu commented Oct 27, 2026

Copy link
Copy Markdown
Contributor Author

Rebased onto 6b1b65d to clear a conflict with #24392 in test/test_security_gate_liveness.py. Main raised _PACKAGE_LINE_BUDGET from 32,613 to 32,616 for the enterprise install-tree write fence, and this PR raised it for SharePoint locator attestation. The resolution keeps both notes, restarts this PR's note from 32,616 and pins the measured total, 32,676 lines (this PR's 60 on top of main's 32,616). No code changed, and the rest of the patch is identical.

@rubencu

rubencu commented Oct 27, 2026

Copy link
Copy Markdown
Contributor Author

Rebased onto cd89e05 as d26784a. This resolves the _PACKAGE_LINE_BUDGET conflict with #23840 in test/test_security_gate_liveness.py. Main's note stays first, and this PR's note now starts from 32,635. The measured package total is 32,695 (main + 60). The rest of the patch is identical to 87b9a41.

@rubencu

rubencu commented Oct 29, 2026

Copy link
Copy Markdown
Contributor Author

Rebased onto main to resolve conflicts. Each resolution keeps both sides:

No code of this PR changed. This PR's tests, the tests main changed since the last push, the composition contracts and the static gates pass. The only failures are ones main shows on this host too.

@rubencu

rubencu commented Oct 29, 2026

Copy link
Copy Markdown
Contributor Author

Pushed 8852bf4, rebased onto main 3cdcfe893c. The rebase resolved one conflict in test/test_security_gate_liveness.py: main raised _PACKAGE_LINE_BUDGET to 32,799, so this PR's budget is now 32,859 (main's value plus this PR's 60 lines), which equals the measured package total. The rest of the patch is unchanged.

On 9fb1d4aa52, the PR's tests, the tests main changed since ca8f35673b, the composition contracts and the static gates passed (9,078 tests); the only failures also fail on main on this host. After the final re-sync onto 3cdcfe893c, main's newly changed test file and the composition contracts pass again.

@rubencu

rubencu commented Oct 29, 2026

Copy link
Copy Markdown
Contributor Author

Rebased onto main 203a9ce3 (head 92079556). One conflict, in test/test_security_gate_liveness.py: main raised the security package line budget to 32,802 (#23855). This PR's own increment is unchanged (60 lines), so the pin is now 32,862, the measured total. Main's budget paragraphs stay whole and this PR's paragraph follows them. No code change.

@rubencu

rubencu commented Nov 5, 2026

Copy link
Copy Markdown
Contributor Author

Rebased onto main 5ac074ef (head 9e885f85). Conflicts, each resolved by keeping both sides:

No behavior change beyond the merge.

@rubencu

rubencu commented Nov 5, 2026

Copy link
Copy Markdown
Contributor Author

Pushed 0d595f4, rebased onto main d4c35f448f. The rebase resolved a conflict in docs/system-specs/modules/security.md: #30279 added a sentence to the ?token= parameter line directly below the SharePoint locator paragraph this PR adds, and both are kept. The PR's own diff is unchanged.

@rubencu

rubencu commented Nov 5, 2026

Copy link
Copy Markdown
Contributor Author

Pushed fc44ffe, rebased onto main 165dbc0eba. The rebase resolved a conflict in test/test_security_gate_liveness.py: #30662 raised _PACKAGE_LINE_BUDGET by one, to 34,524, for the Railway consent host, and this PR raised it from 34,523 by 60 lines. Both are kept: main's comment line stays, the budget is now 34,584, and this PR's comment says it was raised from 34,524. The rest of the PR's diff is unchanged.

@rubencu

rubencu commented Nov 12, 2026

Copy link
Copy Markdown
Contributor Author

Pushed f7b4389, rebased onto main ab65f7add4. The rebase resolved conflicts in eleven files, keeping both sides:

  • Main now hands the display battery and the credential passes a snapshot of the pass-4 token-parameter prefixes, the same way this PR hands them the attested SharePoint locator digests. display_battery, redact_credentials, _document_link_spans, _credential_redaction_plan and _display_redaction_cache_key take both, and every merged call names both. Main also split the credential plan into _main_plan and _later_passes and added the structured-detector path, so the attested digests now pass through those as well, which adds redaction_structured/compat.py to this PR's diff.
  • permission_requests.py keeps main's new mcp_identity_unreadable_reason argument and this PR's comment. The ACP and security specs keep both new sections.
  • Pins: the lazy-import list adds locator_attestations beside main's crewmate_harness_notice (151 names), and the security package budget is 35,779 (main's 35,712 plus this PR's 67 lines, re-measured). One cache-key stub in test_sharepoint_locator_attestations.py passes main's new keyword through. This PR's edits to test_display_verdicts.py dropped out, because main's stubs already pass every keyword through.

Nothing else changed.

@rubencu

rubencu commented Nov 12, 2026

Copy link
Copy Markdown
Contributor Author

Pushed 688a099, rebased onto main d9839eee5d. The rebase resolved a conflict in test/test_security_gate_liveness.py: main raised _PACKAGE_LINE_BUDGET by 92, to 35,804, for paths_resolver, and this PR adds 67 lines. Both are kept: main's comment stays, the budget is now 35,871 (re-measured), and this PR's comment says it was raised from 35,804. The rest of the PR's diff is unchanged.

@KiroCrewCaptain

KiroCrewCaptain commented Nov 12, 2026 •

Copy link
Copy Markdown
Collaborator

Code change approver recommendation: A person should decide

Why: the decider raised 1 possible problem and estimates a 60% chance that this change introduces a defect; a person should check it.

The code looks careful and fails closed: the public policy is empty, the redactor only skips locators that match a digest, and KAS permission handling still sees a KAS-only view of the verdict cache. But the PR does not reach its stated goal. No supplier declares provider fields, so no SharePoint link is kept in any deployment, and the PR itself says enabling it needs a maintainer decision (label: needs-author-decision). It adds a large dormant mechanism (durable registry, boot preload, revalidation threads, identity-verdict recording for every harness) that delivers nothing yet. Request changes: land it with an approved supplier, or get maintainer sign-off first.

Things to check (possible problems the decider raised; at this confidence many turn out to be fine):

  • The stated goal is to keep provider-generated SharePoint item locators from being rewritten as [REDACTED: credential] in rendered messages and persisted history. No supplier exists in this tree, or according to the description in any companion, so no SharePoint link survives redaction any better t (src/kiro_crew/platform/defaults.py, medium)

An automated recommendation, not a review: a maintainer still decides. It approves on its own only when the estimated chance of a defect is at most 12% and the change has at most 400 changed lines. Decider v3c (217fe607) · commit b276e81b52 · 2162 changed lines. If this recommendation is wrong, reply here: replies are how the approver is evaluated.

@rubencu

rubencu commented Nov 13, 2026

Copy link
Copy Markdown
Contributor Author

Pushed 5d9c69e, a conflict-only rebase onto main 52cc80f678. Main's #35833 added mcp_listing_gateway to the dashboard's lazy-import table in test/test_dashboard_server_composition_contract.py, on the same line as this PR's locator_attestations. The table now lists both, and its count pin is 152 (main's 151 plus this PR's 1). No other code, test or description changed.

@rubencu

rubencu commented Nov 13, 2026

Copy link
Copy Markdown
Contributor Author
  • span=0f94af2817ab — fixed

Fixed in 662c9d4: the rootdir conftest's tryfirst teardown now joins every live locator-attestation-load thread, bounded, before any fixture unwinds, beside the consent-worker join, and fails the test if one is still running (conftest._join_locator_attestation_load). test_the_floor_joins_a_background_read_before_this_tests_home_unpins holds a registry read in flight and shows the join waits for it under the test's own KIROCREW_HOME; it fails without the join.

@rubencu

rubencu commented Nov 13, 2026

Copy link
Copy Markdown
Contributor Author

Pushed 662c9d4, rebased onto main 27e9359419.

The First Principles block is the supplier decision recorded earlier and still needs a maintainer. No other code, test or description changed.

@rubencu

rubencu commented Nov 13, 2026

Copy link
Copy Markdown
Contributor Author

Pushed b276e81, rebased onto main 57ebed3246.

No other code, test or description changed.

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

fork Pull request from a fork (external contributor) merge conflict Branch has merge conflicts with its base — author must resolve before merge needs-author-decision PR blocked on author input readiness: action required A blocking check or review needs attention

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants