Skip to content

fix(mcp): list an app's tools only while the app is enabled - #16127

Merged
iamwhatever merged 1 commit into
mainfrom
fix/16099-hide-disabled-app-tools
Oct 9, 2026
Merged

iamwhatever merged 1 commit into
mainfrom
fix/16099-hide-disabled-app-tools

Conversation

@pepmach

@pepmach pepmach commented Oct 2, 2026 •

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: kirocrew-core advertises the tools of Issue Radar, Dev Fleet and Ops Mission Control only while that app is enabled, so a session is never told about a tool whose every call the gateway refuses.

With the Issue Radar, Dev Fleet and Ops Mission Control apps disabled, tools/list on kirocrew-core still carried all eight of their tools (issue_radar_record_investigation, issue_radar_crew_read, issue_radar_crew_record, pod_up, pod_down, pod_status, pod_ls, ops_mission_control_api). Each of those tools reaches a gateway route that refuses the call while its app is disabled (app_not_enabled, #10742), so the model was offered eight capabilities that could only refuse -- and, because kiro-cli reads tools/list once per session and Crew's own servers are exempt from tool_search deferral, each one cost its full JSON schema in every request of every such session.

Why it matters

Every session on an install with these apps off -- the default, since every builtin ships defaultEnabled: false -- pays context for eight unusable tools in every request, and an agent that picks one gets a refusal it then reasons about or retries. The same install with Dev Fleet on must keep exactly the four pod tools, so the fix has to follow enablement per app rather than drop the domain.

Not a goal

  • Changing any tool's name, description or schema: the fix decides WHETHER a descriptor is emitted, never its shape.
  • Changing the call path: dispatch() and every handler are untouched, and the gateway route stays the enforcement point.
  • Surfacing a newly enabled app's tools mid-session: kiro-cli caches tools/list for the session's life, as it does for the computer-use tools and the per-session policy filter.
  • Gating any kirocrew-core tool that belongs to no app.

What changed (motivation → approach → change)

Root cause: kiro_crew.mcp_tools.build_tool_list() concatenated every domain's schemas() unconditionally, while the gateway routes those tools call decide enablement per request through kiro_crew.apps.manager.is_app_enabled. The listing and the call must agree from ONE predicate, so the listing reads the same installed.json through the same module the routes read; the MCP server process reads it directly, as mcp_computer._list_tools already reads the computer-use keystone. The per-session managedToolPolicy filter was rejected as the seam: cached per session with no TTL, and its tools/call leg fails closed.

  • src/kiro_crew/mcp_tools/apps.py: TOOL_APPS maps each app tool to the installed app name its routes gate on, and advertised() returns schemas() minus the tools of every app whose tri-state app_enabled_state is a definite False. An unreadable record (None, or a read that raises) lists the tools: fail-open on listing, because the once-per-session cache would make hiding on a transient fault permanent, while a listed tool whose call the gateway refuses is not a hole. schemas() and HANDLERS keep naming every tool whatever the enablement.
  • src/kiro_crew/mcp_tools/__init__.py: build_tool_list() reads a domain's advertised() in place of schemas() on the full build only. Main's refactor(mcp): names-only in-process tool read; drop the per-builder loop skips #18001 replaced the per-builder asyncio.get_running_loop() skips with a names-only build (build_tool_names(), what in-process discovery reads on the gateway loop); this PR's earlier running-loop guard was that skip written a third time and is gone -- the names-only path takes schemas() and never reads an app's enablement record, so a disabled app's tools still appear by name in the discovery inventory, under its "Declared" badge.
  • Main's feat(design-tweak): scoped MCP tool for comment-thread progress (Fixes #13263) #17622 added a ninth apps-domain tool, design_tweak_update_thread, whose route (the app reverse proxy, handle_app_api_proxy) refuses a disabled Design Tweak with 403 app_not_enabled on the same predicate; it is gated like the other three apps' tools (TOOL_APPS -> design-tweak, pinned to the app's manifest name).
  • docs/architecture/mcp.md (owning spec): the App bridges entry, the two-halves paragraph, and the section "An app's tools follow the app's enablement" recording the predicate, the fail-open rule, the mid-session caveat and the names-only rule. docs/system-specs/modules/app-kit-platform.md: one paragraph under section 7 pointing enablement readers at it.

Backwards compatibility

Compatible: no tool is renamed or reshaped, and an enabled app lists exactly the tools it listed before. The only observable change is that a disabled or uninstalled app's tools are absent from tools/list; a call to one of them was already refused by the gateway and still is, with the 403 body each app's routes produce (Dev Fleet and Design Tweak app_not_enabled, Ops Mission Control app_disabled, Issue Radar no code). Enabling an app mid-session surfaces its tools at the next session, as the computer-use tools behave.

Reader: test/test_mcp_apps_tool_gate.py pins the listing to the route's predicate with real installed.json records (none installed, disabled, one enabled, all enabled, unreadable), the names-only build's zero enablement reads, the hidden tool's dispatch() still reaching the route refusal, and every TOOL_APPS value against the app's own name.
Reader: test/test_mcp_discovery.py::TestManagedToolsInProcessNamesOnly::test_core_names_are_correct_and_complete holds every app enabled while comparing the names-only build to the full build, the one state in which the two now legitimately differ.

Tests

Red-first, -n0: test/test_mcp_apps_tool_gate.py failed 8 of 10 before the fix on 3d465a6066; on this head 16 pass. test/test_mcp_tool_registry.py reads parity from the declaration (schemas()), since build_tool_list() may legitimately omit a declared tool, and test_a_domain_advertises_only_what_it_declares pins advertised() to a subset of schemas(). test/test_mcp_tool_titles.py, test/test_codex_session_mcp.py, test/test_issue_radar_crew_mcp_tools.py and test/test_issue_radar_investigation_tool.py hold enablement open while enumerating kirocrew-core. On this head: ten named suites, 778 passed -n0; mypy src/kiro_crew clean; black ratchet, isort, flake8, docs lint and check_comment_history.py pass; the full backend suite in one -n auto leg, 174401 passed, 76 failed, all 76 reproduced identically on a pristine origin/main checkout in files this diff does not touch.

Edit log: 04c8253cf9 initial; 24be19330f comment lines restated (Comment History Gate); 465b71d428 read guard, two-reader contract, refusal shapes; 29c5a00be8 running-loop guard (since removed); 215c326e3f, 3c7e9033d7, 4711441ab5 rebases; 461f5acd38 rebase, #18001 adaptation, #17622 tool gated.

Manual verification

Round 1 drove the real stdio server (python -m kiro_crew mcp-core, initialize + tools/list): no app installed lists none of the app tools; dev-fleet enabled lists exactly the four pod tools. No UI change.

Screenshots / video

None: no visual change (backend listing only).

Related Issues

Fixes #16099

Pattern harvest

Rule candidate: review-prompt
Pattern: a server advertises a capability whose call path is gated elsewhere (an app enable, a keystone, a policy) without reading the same gate; listing and call must agree from one predicate.

Checklist

  • At most two commits (one is the norm), with a Conventional Commits title (feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)
  • Existing tests pass and new tests added for new functionality
  • Self-review completed; code follows project style guidelines
  • Documentation updated (if applicable)
  • No secrets, credentials, or internal references in the diff

Contribution License Agreement

@pepmach
pepmach requested a review from a team as a code owner October 2, 2026 09:42
@pepmach
pepmach requested a review from bolichen97 October 2, 2026 09:42
@github-actions github-actions Bot added the readiness: checking Automated validation is still running label Oct 2, 2026
@github-actions

github-actions Bot commented Oct 2, 2026 •

Copy link
Copy Markdown
Contributor

Design Review (Opus 5.5) — ✅ PASS

Design-level review of 461f5acd386b7dece7869894d32b96c9d0e8594e — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

Design-Verdict: PASS

Check that design_tweak_update_thread belongs in the gate. The diff hides nine tools across four apps, but the frozen Goal and the Problem section name eight tools across three.

Watch

  • The TOOL_APPS hunk adds "design_tweak_update_thread": "design-tweak". The Goal names only "Issue Radar, Dev Fleet and Ops Mission Control", and the Problem section lists eight tools without it. The docs in this patch do say "nine". That means the Design Tweak gate depends on its route being refused through apps/routes.py's app_not_enabled proxy check, and the description never claims or proves that.
    Clears when: a PR comment or the Tests section names Design Tweak's refusal path and the test that covers it, or the entry is moved to its own PR.

[DESIGN-REVIEWED] 461f5ac

Description read

[DESCRIPTION-READ] b597de6942378eebeb8511efff1e36c24d1d167c24bd37556157bae5d3b20b70

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 Oct 2, 2026 •

Copy link
Copy Markdown
Contributor

GPT 6.1 Review — ✅ no blocking findings

GPT 6.1 completed its review of 461f5acd386b7dece7869894d32b96c9d0e8594e and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] 461f5ac

False positive or not applicable? A repository writer can comment:
/ai-review override gpt 461f5acd386b7dece7869894d32b96c9d0e8594e: <one-sentence reason>

@github-actions

github-actions Bot commented Oct 2, 2026 •

Copy link
Copy Markdown
Contributor

First Principles Review (Opus 5.5) — ✅ PASS

Premise-level review of 461f5acd386b7dece7869894d32b96c9d0e8594e — 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: PASS

Check that hiding design_tweak_update_thread is intended: the description says "eight tools", but TOOL_APPS gates nine.

Not justified as shipped

  • Item 4 — undeclared: the description says "TOOL_APPS maps each of the eight tools", but the diff also maps "design_tweak_update_thread": "design-tweak". Its proxy route refuses the call with app_not_enabled (apps/routes.py:4380), so the harm is the same one and it carries no premise risk.

What this change ships

Inventory (6 items) — 5 justified

Intent: stop a session from being offered app tools that can only be refused while their app is off. This is a FIX.

  1. With Issue Radar disabled or not installed, its three tools drop out of kirocrew-core's tools/list — justified
  2. With Dev Fleet disabled, the four pod_* tools drop out of the listing — justified
  3. With Ops Mission Control disabled, ops_mission_control_api drops out of the listing — justified
  4. With Design Tweak disabled, design_tweak_update_thread drops out of the listing — undeclared (the description counts eight tools)
  5. If an app's enablement record can't be read, its tools stay listed (fail-open) — justified
  6. A domain can now define an advertised() hook that build_tool_list reads, and the names-only discovery path skips it — justified

[FIRST-PRINCIPLES-REVIEWED] 461f5ac

Description read

[DESCRIPTION-READ] b627c907420f9347fc19ab254ae6e84706a836517863916ce363c13ca9e5604e

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 added readiness: action required A blocking check or review needs attention and removed readiness: checking Automated validation is still running labels Oct 2, 2026
@github-actions

github-actions Bot commented Oct 2, 2026 •

Copy link
Copy Markdown
Contributor

Opus 5.5 Review — ✅ no blocking findings

Reviewed 461f5acd386b7dece7869894d32b96c9d0e8594e — this comment is updated in place on each push.

Review details

Disabling an app now makes any later transparent respawn of the shared kirocrew-core backend fail for sessions that were shown that app's tools.

FINDING — src/kiro_crew/mcp_tools/apps.py:306 — A session starts while an app is enabled, so it holds that app's tools, such as pod_up through pod_ls. The operator then turns the app off. When the pooled backend later dies mid-call, the replacement's tools/list runs build_tool_list() → advertised(), which applies if state is False: hidden.update(...). In replacement.py, describe_surface_change then reports gone=pod_down,..., and _refuse_replacement gives the in-flight call a terminal error and forces a reconnect. Before this change, the respawn was adopted without the session noticing. Turning an app off is an ordinary dashboard action, and the documented behaviour, "disabling one mid-session leaves them listed and refusing", does not hold across a respawn → Fix: when comparing surfaces during a respawn, don't count a tool listed in TOOL_APPS that is declared but hidden as gone, for example by comparing against the declared set. That change goes in untouched code (tool_surface.py / daemon/replacement.py). Otherwise, document this interaction in mcp.md.

[OPUS-REVIEWED] 461f5ac

Verdict parsed from the review's SHA-scoped output markers for commit 461f5acd386b7dece7869894d32b96c9d0e8594e.

False positive or not applicable? A repository writer can comment:
/ai-review override fable 461f5acd386b7dece7869894d32b96c9d0e8594e: <one-sentence reason>

@pepmach
pepmach force-pushed the fix/16099-hide-disabled-app-tools branch from 04c8253 to 24be193 Compare October 2, 2026 10:44
@github-actions github-actions Bot added readiness: checking Automated validation is still running readiness: action required A blocking check or review needs attention and removed readiness: action required A blocking check or review needs attention readiness: checking Automated validation is still running labels Oct 2, 2026
@pepmach
pepmach force-pushed the fix/16099-hide-disabled-app-tools branch from 24be193 to 465b71d Compare October 2, 2026 12:44
@github-actions github-actions Bot added readiness: checking Automated validation is still running readiness: action required A blocking check or review needs attention and removed readiness: action required A blocking check or review needs attention readiness: checking Automated validation is still running labels Oct 2, 2026
@pepmach
pepmach force-pushed the fix/16099-hide-disabled-app-tools branch from 465b71d to 29c5a00 Compare October 2, 2026 13:44
@github-actions github-actions Bot added readiness: checking Automated validation is still running readiness: action required A blocking check or review needs attention and removed readiness: action required A blocking check or review needs attention readiness: checking Automated validation is still running labels Oct 2, 2026
@github-actions github-actions Bot added readiness: passed Eligible automated validation passed for the current revision and removed readiness: action required A blocking check or review needs attention labels Oct 2, 2026
@pepmach

pepmach commented Oct 2, 2026

Copy link
Copy Markdown
Contributor Author

CI on 29c5a00be80f, two reds, neither from this diff, both green on one --failed rerun: GPT 5.6 Review was a lane fault, not a finding — the model stream disconnected after five reconnects and the gate failed closed with no [GPT-REVIEWED] marker (#16166); Backend Tests (Windows) (3) failed test/test_discord.py::TestDispatcher::test_a_dm_turn_opens_the_crew_log_the_work_ledger_writes_into on OSError: could not acquire exclusive file lock … 0.00s, the same crew-log lock taken on the event loop as #16093 (the same OSError as #15676), in a file this diff does not touch — main a34299dec7 (#16102, "touch the crew log off the event loop in every async test") may retire that class. Coverage Gate was downstream of that shard.

@iamwhatever

Copy link
Copy Markdown
Contributor

Reviewed for merge triage, not approving from my side: it changes which MCP app tools are listed and therefore callable (an authorization surface: tools should vanish while an app is disabled). That needs a security owner's sign-off. CI is green.

@github-actions github-actions Bot added readiness: checking Automated validation is still running readiness: passed Eligible automated validation passed for the current revision and removed readiness: action required A blocking check or review needs attention readiness: checking Automated validation is still running labels Oct 6, 2026
@iamwhatever

Copy link
Copy Markdown
Contributor

The commits since the hold (rounds 5 to 7) are rebases with no change of substance, so the code is the same one the hold was about. One correction to my earlier comment: this PR does not change whether a tool can be called. dispatch() and the handlers are untouched, a hidden tool still reaches its route, and the gateway's is_app_enabled check is still what refuses the call. What it does change is which tools kirocrew-core lists. It reads installed.json from inside the MCP server process, lists the tools if that read fails or runs on an event loop, and adds a new hook in build_tool_list(). Those choices about what gets exposed should still be confirmed by a security owner. @bolichen97 is still a requested reviewer and has not reviewed yet. Once that sign-off lands, I see no remaining blockers. CI readiness is passed and the branch is mergeable.

@pepmach

pepmach commented Oct 6, 2026

Copy link
Copy Markdown
Contributor Author

Thanks, that correction matches the change: dispatch() and the handlers are untouched, a hidden tool still reaches its route and gets the route's 403 refusal (Dev Fleet app_not_enabled, Ops Mission Control app_disabled, Issue Radar no code -- pinned by test/test_mcp_apps_tool_gate.py), and only build_tool_list() changes, through advertised() in src/kiro_crew/mcp_tools/apps.py. The fail-open rule is deliberate and recorded in docs/architecture/mcp.md ("An app's tools follow the app's enablement"): a tool is withheld only when the app's app_enabled_state is a definite False; an unreadable or absent installed.json lists everything, because the gateway's is_app_enabled check still refuses the call, so a read failure can hide tools but never expose a call. @bolichen97 remains the requested reviewer for that choice; nothing further is planned on this branch before that review.

@pepmach

pepmach commented Oct 6, 2026

Copy link
Copy Markdown
Contributor Author

One more of the three choices, for the same review: advertised() skips the installed.json read entirely when an event loop is running. A running loop means the caller is not the stdio server but the gateway's in-process discovery probe (mcp_discovery._managed_tools_in_process, calling _list_tools() from probe_server), which keeps only tool names, so the full declaration is returned there and no file read is charged to the gateway's loop -- the same rule mcp_tools/spawn.py::_agent_roster_hint and mcp_tools/control.py already apply for that caller. Recorded in docs/architecture/mcp.md ("On the gateway's event loop the gate is skipped, not charged to the loop").

@github-actions github-actions Bot added the merge conflict Branch has merge conflicts with its base — author must resolve before merge label Oct 7, 2026
kirocrew-core advertised the eight tools of Issue Radar, Dev Fleet and
Ops Mission Control whatever the apps' state, while every route those
tools call refuses with app_not_enabled when its app is disabled
(#10742). A session with the apps off was told about tools that could
only refuse, and paid each one's full schema in every request, since
kiro-cli reads tools/list once per session and Crew's own servers are
exempt from tool_search deferral.

The apps domain now maps each tool to the installed app behind it
(TOOL_APPS) and emits a descriptor only while that app is enabled,
read through the same kiro_crew.apps.manager installed.json reader the
routes use, so listing and call agree from one predicate. Hiding needs
a definite False; an unreadable record lists, as the per-session policy
filter does, because the once-per-session cache would make hiding on a
transient fault permanent while a listed tool that refuses is not a
hole. schemas() and HANDLERS keep declaring all eight, so the registry
parity tests read the declaration; build_tool_list() reads a domain's
advertised() in place of schemas() when it defines one.

Fixes #16099
@pepmach
pepmach force-pushed the fix/16099-hide-disabled-app-tools branch from 4711441 to 461f5ac Compare October 8, 2026 23:57
@github-actions github-actions Bot added readiness: checking Automated validation is still running and removed readiness: passed Eligible automated validation passed for the current revision labels Oct 8, 2026
@github-actions

github-actions Bot commented Oct 8, 2026

Copy link
Copy Markdown
Contributor

UX Review (Opus 5.5) — ⏭️ skipped

Revision 461f5acd386b7dece7869894d32b96c9d0e8594e touches no user-facing surface (no changes under website/ or committed screenshots), so the UX review was skipped. Advisory — does not block merge.

@github-actions github-actions Bot added readiness: passed Eligible automated validation passed for the current revision readiness: checking Automated validation is still running and removed readiness: checking Automated validation is still running readiness: passed Eligible automated validation passed for the current revision merge conflict Branch has merge conflicts with its base — author must resolve before merge labels Oct 9, 2026
@pepmach

pepmach commented Oct 9, 2026

Copy link
Copy Markdown
Contributor Author
  • span=b6e03c218403 -- accepted: the description now names Design Tweak's refusal path and the test that covers it

The item was read against the description as it stood at 00:01Z; the description was rewritten at 00:44Z and 00:49Z (this lane does not re-run on a body edit). ## What changed, third bullet, now says that #17622's design_tweak_update_thread reaches Design Tweak through the app reverse proxy (handle_app_api_proxy in src/kiro_crew/apps/routes.py), which refuses a disabled app with 403 app_not_enabled on the same is_app_enabled predicate as the other three apps, and that the tool is gated like theirs via TOOL_APPS -> design-tweak, pinned to the app's manifest name.
## Backwards compatibility names the covering test in its first Reader: line: test/test_mcp_apps_tool_gate.py pins every TOOL_APPS value against the app's own name (the Design Tweak entry against its manifest), lists the ninth tool only with its app enabled (test_one_enabled_app_lists_exactly_its_tools is parametrized over the four apps), and keeps schemas(), HANDLERS and TOOL_APPS on the same nine names. ## Tests records that file at 16 passing on this head.
The **Goal:** line and ## Problem / Motivation are kept as the owner wrote them: they describe the defect as filed in #16099, when the domain had eight tools; the ninth arrived on main with #17622 after round 7, which is why the gate grew by one entry in the same change rather than in a PR of its own.

@pepmach

pepmach commented Oct 9, 2026

Copy link
Copy Markdown
Contributor Author
  • span=b1d4b23fe11d -- accepted: item 4 is now declared in the description

The item was read against the description as it stood at 00:02Z; it was rewritten at 00:44Z and 00:49Z (this lane does not re-run on a body edit). The "maps each of the eight tools" sentence is gone: ## What changed now says TOOL_APPS maps each app tool to the installed app name its routes gate on, and its third bullet declares the ninth entry by name -- #17622's design_tweak_update_thread, whose reverse-proxy route (handle_app_api_proxy) refuses a disabled Design Tweak with 403 app_not_enabled on the same predicate -- as gated like the other three apps' tools, which is the item's own reading of the harm.
## Backwards compatibility now lists Design Tweak among the refusal shapes (Dev Fleet and Design Tweak app_not_enabled``), and the Reader: line points at `test/test_mcp_apps_tool_gate.py`, where the entry is pinned to the app's manifest name and listed only while the app is enabled.

@pepmach

pepmach commented Oct 9, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=0b188f53f0fa -- rebutted: a replacement whose listing lacks a now-disabled app's tools is a real schema move, and refusing it is the gateway's stated contract, not a defect this change introduces

The sequence is read correctly: a session served pod_* while Dev Fleet was enabled, the operator disables the app, the pooled backend dies mid-call, and the replacement's tools/list omits the four tools, so tool_surface.describe_surface_change reports them gone and _refuse_replacement answers the in-flight call with a terminal error and the stub reconnects. That path exists for exactly this case. src/kiro_crew/mcp_gateway/daemon/replacement.py (_refuse_replacement, lines 71-104): "the frozen tool set lives in the client and no gateway-side write can refresh it, so the only honest options are to adopt a process whose schema the session cannot see has moved, or to refuse and say why. Refusing lands on the give-up path ... a stated failure at a call boundary instead of a wrong one." The replacement's listing has genuinely moved -- the operator removed the app -- and the client message ("this session's tools are stale -- reconnect to pick up the new ones") is literally true: after the reconnect the session lists what the gateway will serve.
Hiding the difference, as proposed (comparing the replacement against the declared set in tool_surface.py / replacement.py), would make the gateway adopt a replacement under a listing the session cannot have, which is the outcome that module refuses by design; it would also special-case this PR's TOOL_APPS inside the gateway daemon, code this change does not touch. The pre-change behaviour (adopt silently, keep listing eight tools that the route refuses) is the defect #16099 names; a failed in-flight call on the crash-plus-disable path, followed by a correct listing, is the recovery the daemon chose for every schema move. docs/architecture/mcp.md records the mid-session rule this is an instance of: tools follow enablement at the next listing, never inside a running session.

@iamwhatever iamwhatever left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This PR stops kirocrew-core from advertising the tools of a disabled built-in app (Issue Radar, Dev Fleet, Ops Mission Control, Design Tweak). Every call to those tools is refused by the app's gateway route anyway, so hiding them fixes the cause: the model being offered capabilities that can only refuse. I read the production diff (mcp_tools/__init__.py, mcp_tools/apps.py) and the tests. advertised() hides a tool only when app_enabled_state returns a readable False, and fails open on None or an exception. The names-only discovery path never reads enablement, and the route stays the enforcement point, so this is listing hygiene, not a security change. test_mcp_apps_tool_gate.py pins TOOL_APPS against each app's real name. Checks are green: the one PR Hygiene failure has a newer SUCCESS. All lanes PASS, and the Design and First Principles items and the Opus advisory have dispositions on this head. One non-blocking note: as Opus found, if an app is disabled and the shared backend later respawns, sessions that were shown that app's tools take one terminal "tools are stale — reconnect" error. The author's rebuttal is reasonable since that is daemon/replacement.py's stated contract, but a sentence on it in docs/architecture/mcp.md would help operators. — crew-pr-reviewer

@iamwhatever
iamwhatever enabled auto-merge (squash) October 9, 2026 06:06
@iamwhatever
iamwhatever merged commit 9c93465 into main Oct 9, 2026
121 of 122 checks passed
@iamwhatever
iamwhatever deleted the fix/16099-hide-disabled-app-tools branch October 9, 2026 06:07
@github-actions github-actions Bot removed the readiness: passed Eligible automated validation passed for the current revision label Oct 9, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

kirocrew-core lists the tools of disabled Issue Radar, Dev Fleet and Ops Mission Control apps

2 participants