release: promote dev to master (v1.0.0-beta.48, sync branch)

[jaylfc]

Replaces #2365 (dev->master direct hit the strict up-to-date rule: master carries 3 Dependabot merges dev never got - h2 4.4.1, aiohttp 3.14.3, npm group). Same pattern as sync/dev-to-master-beta45 and release/promote-beta46: branch from dev + merge master, so the promoted tree is dev's content plus master's newer lockfile entries. Version lines verified intact after the lock merge (uv.lock 1.0.0b48, package.json 1.0.0-beta.48; test_version_lock_sync passes).

After this merges: back-merge master into dev to restore tree identity (diff dev master must be empty), then tag v1.0.0-beta.48 and cut the GitHub Release.

Closes out the tsk-laezfg / #2362 release-time verification via the secret-ignores-gate run on this PR and the master push.

Summary by CodeRabbit

• New Features
  • Added five Hailo-10H model options with hardware-aware catalog support.
  • Added editable agent handles for authorized users.
  • Added wallpaper fit controls with per-device persistence.
  • Added task quarantine and lead-only unquarantine workflows.
  • Added agent-loop status visibility and queued chat handling.
  • Added chat export support and external model API access.
• Bug Fixes
  • Improved session security, authorization privacy, memory-path validation, project reordering, and model context-window handling.
• Chores
  • Updated the release to beta.48 and upgraded SPA builds to Node.js 22.
• Documentation
  • Expanded release, model, agent, and security guidance.

[qodo-code-review]

ⓘ Qodo reviews are paused because the subscription is no longer active. Ask your workspace admin to reactivate the subscription to resume reviews. Manage billing

[github-actions]

👋 Thanks for the PR! This one targets master, which is our
stable branch (it's what live installs track). Please retarget it to
dev — click Edit next to the PR title and change the base
branch dropdown from master to dev. Your commits and any review
carry over, nothing is lost.

See CONTRIBUTING.md for the branch model.

[coderabbitai]

📝 Walkthrough

Walkthrough

This release updates agent turn handling, task quarantine, Hailo model support, desktop wallpaper settings, authorization behavior, CI safeguards, storage consistency, documentation, and regression coverage.

Changes

Platform and release updates

Layer / File(s)	Summary
Release, catalog, and governance CHANGELOG.md, README.md, app-catalog/models/*, docs/*, pyproject.toml, tinyagentos/registry.py	Adds beta.48 metadata, five Hailo HEF manifests, context-window support, governance documentation, and updated agent manuals.
CI validation gates .github/workflows/*, scripts/check_secret_ignores.py, scripts/check_store_wiring.py, .gitignore	Adds secret-ignore and store-wiring gates, their scripts, tests, workflows, and required ignore coverage.

Desktop and agent behavior

Layer / File(s)	Summary
Desktop wallpaper and registry controls desktop/src/stores/theme-store.ts, desktop/src/theme/tokens.css, desktop/src/App.tsx, desktop/src/components/Desktop.tsx, desktop/src/apps/SettingsApp.tsx, desktop/src/apps/agents/RegistryPanel.tsx	Adds per-device wallpaper-fit persistence, fit-mode rendering, settings controls, and owner/admin registry-handle editing.
AgentLoop turn coordination tinyagentos/agent_loop.py, tinyagentos/agent_chat_router.py, tinyagentos/routes/taos_agent.py	Adds queued per-agent turns, safe-point processing, supervised subagents, progress handling, status reporting, and stream cleanup.
Chat export and storage consistency tinyagentos/chat/chat_exporter.py, tinyagentos/projects/lists_store.py, tinyagentos/registry.py	Adds deterministic chat export with size handling and strengthens list reorder rollback and manifest context-window loading.

Task and route behavior

Layer / File(s)	Summary
Task strikes and quarantine lifecycle tinyagentos/projects/strike_store.py, tinyagentos/projects/task_store.py, tinyagentos/app.py, tinyagentos/routes/projects.py	Adds strike persistence, task quarantine and unquarantine operations, strike metadata, lifecycle wiring, audit events, and lead-only unquarantine handling.
Authentication and authorization controls tinyagentos/auth_middleware.py, tinyagentos/routes/auth.py, tinyagentos/routes/*, tests/test_agent_registry.py, tests/test_agent_scope_requests.py	Binds session checks to User-Agent values, exposes selected model routes, returns existence-hiding 404 responses, and rejects unsafe memory paths.

Sequence Diagram(s)

sequenceDiagram
  participant Client
  participant taos_agent
  participant AgentLoop
  participant ACP
  Client->>taos_agent: Submit chat request
  taos_agent->>AgentLoop: Accept turn or queue message
  AgentLoop->>ACP: Drive active turn
  ACP-->>AgentLoop: Stream response and completion
  AgentLoop->>AgentLoop: Reach safe point and drain queue
  AgentLoop-->>taos_agent: Return frames and status
  taos_agent-->>Client: Stream response frames

Loading

Estimated code review effort: 5 (Critical) | ~120 minutes

Possibly related PRs

• jaylfc/taOS#2333: Adds the same strike-store, task-store wiring, lifecycle integration, unquarantine route, and related tests.
• jaylfc/taOS#2335: Adds the same AgentLoop implementation, design document, and tests.
• jaylfc/taOS#2338: Adds the same five Hailo-10H HEF model manifests.

Suggested reviewers: hognek

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 37.38% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly identifies the release promotion from dev to master and includes the release version and synchronization context.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches 💡 1

📝 Generate docstrings 💡

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch sync/dev-to-master-beta48

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[gitar-bot]

Important

You are using the Gitar free plan. Upgrade to unlock code review, CI analysis, auto-apply, custom automations, and more.

_Gitar

[kilo-code-bot]

Code Review Summary

Status: 9 Issues Found | Recommendation: Address before merge

Overview

Severity	Count
WARNING	8
SUGGESTION	1

Issue Details (click to expand)

WARNING

File	Line	Issue
tinyagentos/routes/agent_auth_requests.py	966	except HTTPException only catches FastAPI HTTP errors from check_agent_identity; non-HTTP exceptions (provider outage, CancelledError) escape as 500s, partially defeating the existence-hiding 404 invariant
tinyagentos/projects/lists_store.py	274	except BaseException: in reorder_entries catches SystemExit/KeyboardInterrupt; use (Exception, asyncio.CancelledError) instead
tinyagentos/projects/task_store.py	453	quarantine_task uses post-commit get_task with no None guard; a concurrent delete yields None, silently dropping the event and writing an empty project_id to the audit log
tinyagentos/routes/agent_auth_requests.py	1087	approve_scope_request returns 404 for non-owners (was 403); breaking API contract for clients keyed on 403
tinyagentos/routes/agent_auth_requests.py	1213	deny_scope_request returns 404 for non-owners (was 403); same breaking API contract concern
tinyagentos/routes/agent_registry.py	679	revoke_registry_entry returns 404 for non-owners (was 403); same breaking API contract concern
tinyagentos/routes/agent_registry.py	790	rotate_tokens returns 404 for non-owners (was 403); same breaking API contract concern
tinyagentos/routes/agent_registry.py	856	update_org_fields returns 404 for non-owners (was 403); same breaking API contract concern

SUGGESTION

File	Line	Issue
tinyagentos/projects/task_store.py	491	unquarantine_task hardcodes from_status="quarantined" for the audit log; safer to derive from existing.get("status") so a future status-filter change doesn't silently write a stale value

Files Reviewed (13 files)

• tinyagentos/routes/agent_auth_requests.py - 3 issues (404 existence-hiding pattern + exception handler)
• tinyagentos/routes/agent_registry.py - 4 issues (404 existence-hiding pattern)
• tinyagentos/projects/lists_store.py - 1 issue (overly broad BaseException catch)
• tinyagentos/projects/task_store.py - 2 issues (post-commit None guard, hardcoded status)
• tinyagentos/routes/taos_agent.py
• tinyagentos/routes/auth.py
• tinyagentos/routes/canvas.py
• tinyagentos/routes/channel_hub.py
• tinyagentos/routes/chat.py
• tinyagentos/routes/memory.py
• tinyagentos/routes/terminal.py
• tinyagentos/cluster/capabilities.py
• tinyagentos/projects/ids.py

Fix these issues in Kilo Cloud

---

_{Reviewed by step-3.7-flash · Input: 75.1K · Output: 16.5K · Cached: 379.4K}

[coderabbitai]

Actionable comments posted: 15

Note

Due to the large number of review comments, Critical, Major severity comments were prioritized as inline comments.

🟡 Minor comments (17)

tests/conftest.py-359-362 (1)

359-362: 🩺 Stability & Availability | 🟡 Minor | ⚡ Quick win

Close task_strikes during fixture teardown.

The fixture initializes this SQLite connection, but teardown never closes it. Repeated client fixtures leak database worker connections.

Add await task_strikes.close() after await project_task_store.close().

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@tests/conftest.py` around lines 359 - 362, Update the client fixture teardown
to close task_strikes after project_task_store.close() by awaiting
task_strikes.close(), ensuring the initialized SQLite connection is released
before teardown completes.

tinyagentos/projects/strike_store.py-21-23 (1)

21-23: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win

Set zip() strictness explicitly.

Line 23 triggers Ruff B905. Use strict=True because _row() requires exactly the six selected columns.

Proposed fix

-    return dict(zip(keys, r))
+    return dict(zip(keys, r, strict=True))

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@tinyagentos/projects/strike_store.py` around lines 21 - 23, Update the _row
function’s dict(zip(...)) call to pass strict=True, ensuring the six expected
keys must match exactly six selected row values.

Source: Linters/SAST tools

tests/test_routes_mcp.py-43-44 (1)

43-44: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win

Fix the Ruff RUF059 warnings.

app is unused at Line 44 and Line 92. Rename it to _app.

Proposed fix

-        client, app = app_client
+        client, _app = app_client
...
-        client, app = app_client
+        client, _app = app_client

Also applies to: 91-92

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@tests/test_routes_mcp.py` around lines 43 - 44, Rename the unused app
variable to _app in both test_list_servers_empty and the corresponding test
around lines 91–92, while preserving the existing app_client unpacking and test
behavior.

Source: Linters/SAST tools

desktop/src/theme/tokens.css-206-223 (1)

206-223: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win

Reset repetition for every non-tile fit.

After a user selects tile, its attribute selector keeps background-repeat: repeat. Selecting fit or center does not restore no-repeat, so the image can repeat in letterbox space. Set background-repeat: no-repeat for all non-tile fit selectors.

Proposed fix

 .taos-wallpaper[data-wallpaper-fit="fill"] {
   background-size: cover;
+  background-repeat: no-repeat;
 }
 .taos-wallpaper[data-wallpaper-fit="fit"] {
   background-size: contain;
+  background-repeat: no-repeat;
 }
 .taos-wallpaper[data-wallpaper-fit="stretch"] {
   background-size: 100% 100%;
+  background-repeat: no-repeat;
 }
 .taos-wallpaper[data-wallpaper-fit="center"] {
   background-size: auto;
   background-position: center;
+  background-repeat: no-repeat;
 }

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@desktop/src/theme/tokens.css` around lines 206 - 223, Update the non-tile
selectors in the wallpaper fit rules—fill, fit, stretch, and center—to
explicitly set background-repeat to no-repeat, while preserving repeat only for
the tile selector.

desktop/src/stores/__tests__/wallpaper-fit.test.ts-62-69 (1)

62-69: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win

Test the Zustand store initializer.

This test calls loadWallpaperFit() directly. It does not verify that a fresh theme-store module initializes wallpaperFit from localStorage. Prepopulate localStorage, call vi.resetModules(), dynamically import theme-store, and assert useThemeStore.getState().wallpaperFit.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@desktop/src/stores/__tests__/wallpaper-fit.test.ts` around lines 62 - 69,
Update the “loads the persisted value as the initial store state” test to verify
the Zustand initializer rather than calling loadWallpaperFit directly:
prepopulate localStorage, reset modules with vi.resetModules(), dynamically
import theme-store, and assert useThemeStore.getState().wallpaperFit equals the
persisted value.

desktop/src/stores/theme-store.ts-196-203 (1)

196-203: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win

Use randomId() in getDeviceId(). Plain HTTP deployments do not provide crypto.randomUUID(). The caught exception makes the preference default to "fill" and prevents persistence. Add a regression test for the missing-randomUUID path.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@desktop/src/stores/theme-store.ts` around lines 196 - 203, Update
getDeviceId() to generate new identifiers with the existing randomId() helper
instead of crypto.randomUUID(), preserving localStorage persistence and
returning the stored identifier. Add a regression test covering environments
where crypto.randomUUID is unavailable.

.claude/skills/taos-development-skill/SKILL.md-451-453 (1)

451-453: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win

Add a language identifier to the fenced block.

Line 451 opens a fenced code block without a language identifier. This violates markdownlint rule MD040.

Proposed fix

-```
+```text
 Store-Unwired-Intentionally: <ClassName>, <why>

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In @.claude/skills/taos-development-skill/SKILL.md around lines 451 - 453,
Update the fenced block containing the Store-Unwired-Intentionally example to
specify the text language identifier, using the existing block content
unchanged.

Source: Linters/SAST tools

tests/test_chat_exporter.py-227-239 (1)

227-239: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win

The UPDATE does not reliably target the message just inserted.

The subquery orders by created_at DESC LIMIT 1. After the first iteration, rows already share ts, so SQLite can select any of the tied rows. The loop can then rewrite the same row repeatedly and leave a later message with its original timestamp. The test still passes, because it only asserts determinism and a count of 3, but it does not create the equal-timestamp scenario it targets. Update by the returned message id instead.

💚 Proposed test change

     for i in range(3):
-        await store.send_message(
+        msg = await store.send_message(
             channel_id="ch1",
             author_id="user1",
             author_type="user",
             content=f"msg{i}",
             content_blocks=[{"type": "paragraph", "text": f"msg{i}"}],
         )
         await store._db.execute(
-            "UPDATE chat_messages SET created_at = ? WHERE id = (SELECT id FROM chat_messages ORDER BY created_at DESC LIMIT 1)",
-            (ts,),
+            "UPDATE chat_messages SET created_at = ? WHERE id = ?",
+            (ts, msg["id"]),
         )
         await store._db.commit()

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@tests/test_chat_exporter.py` around lines 227 - 239, Update the test loop
around store.send_message to capture the returned message identifier, then use
that id in the chat_messages UPDATE instead of selecting the latest row by
created_at. Keep assigning ts and committing each inserted message so the test
reliably creates the intended equal-timestamp scenario.

tinyagentos/chat/chat_exporter.py-188-191 (1)

188-191: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win

A whitespace-only text block aborts the whole channel export.

flatten_body returns the whitespace for a block such as {"type": "paragraph", "text": " "}, so body is truthy and the content fallback is skipped. The guard then raises ChatExportError and export_channel fails for the entire channel. This contradicts the stated design that a block must never brick the export. Consider falling back to content, or to a placeholder, before raising.

🛠️ Proposed fix

-        if not body and msg.get("content"):
+        if not body.strip() and msg.get("content"):
             body = str(msg["content"])
 
         if content_blocks and not body.strip():
-            raise ChatExportError(
-                f"message {msg.get('id')!r} has content_blocks but empty body"
-            )
+            body = f"[{len(content_blocks)} block(s), no text]"

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@tinyagentos/chat/chat_exporter.py` around lines 188 - 191, Update the
message-body handling around flatten_body so whitespace-only flattened text does
not trigger ChatExportError or abort export_channel. Treat body as unusable when
it contains only whitespace, then fall back to the existing content value or an
appropriate placeholder before the content_blocks empty-body guard; preserve the
error only when no usable body or fallback exists.

tinyagentos/agent_loop.py-436-436 (1)

436-436: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win

Unused done result from asyncio.wait in two methods. Both call sites unpack done, pending but read only pending, which Ruff reports as RUF059.

• tinyagentos/agent_loop.py#L436-L436: rename done to _done in await_subagent.
• tinyagentos/agent_loop.py#L481-L481: rename done to _done in await_all_subagents.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@tinyagentos/agent_loop.py` at line 436, Rename the unused done result to
_done in both await_subagent and await_all_subagents when unpacking asyncio.wait
results. Apply the change at tinyagentos/agent_loop.py lines 436-436 and
481-481; leave pending handling unchanged.

Source: Linters/SAST tools

CHANGELOG.md-118-122 (1)

118-122: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win

Describe the old Node 20 state in past tense.

Line 54 states that spa-build runs on Node 22. Line 122 states that CI pins Node 20. Change Line 122 to describe the Dependabot failure before this release changed CI.

Proposed fix

-  spa-build (jsdom 30 requires Node >=22.13; CI pins Node 20).
+  spa-build (jsdom 30 requires Node >=22.13; CI then pinned Node 20).

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@CHANGELOG.md` around lines 118 - 122, Update the changelog entry describing
the grouped jsdom 30 Dependabot failure to use past tense for the former Node 20
CI configuration, while preserving the details about jsdom’s Node 22.13
requirement and the spa-build context.

docs/design/radio-studio.md-325-328 (1)

325-328: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win

Remove Phase 2 commands from the Phase 1 chat contract.

Phase 1 excludes targeted sweeps and audio capture, but the chat contract advertises /stare and /capture. The UI and agent would promise operations that do not ship. Remove these commands or mark them unavailable until the corresponding skills and routes exist.

This finding uses the supplied Phase 1 cut and chat command list.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@docs/design/radio-studio.md` around lines 325 - 328, Update the “Agent Chat”
command list in the Phase 1 contract to remove or explicitly mark `/stare` and
`/capture` as unavailable, while retaining the supported `/survey` and `/alert`
commands.

docs/design/agent-loop-subagents.md-64-64 (1)

64-64: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win

Add language identifiers to all reported Markdown fences.

markdownlint MD040 flags each opening fence because it has no language identifier.

• docs/design/agent-loop-subagents.md#L64-L64: use a text fence and remove the standalone text marker from the block body.
• docs/design/radio-studio.md#L100-L100: use a bash fence.
• docs/design/radio-studio.md#L106-L106: use a bash fence.
• docs/design/radio-studio.md#L265-L265: use a text fence.
• docs/design/radio-studio.md#L281-L281: use a text fence.
• docs/design/radio-studio.md#L334-L334: use a text fence.
• docs/design/radio-studio.md#L423-L423: use a text fence.
• docs/design/radio-studio.md#L451-L451: use a text fence.

This finding is based on the supplied markdownlint results.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@docs/design/agent-loop-subagents.md` at line 64, Update all listed Markdown
fences to include the required language identifiers: use a text fence at
docs/design/agent-loop-subagents.md:64-64 and remove the standalone text marker
from its block body; use bash fences at docs/design/radio-studio.md:100-100 and
106-106; and use text fences at docs/design/radio-studio.md:265-265, 281-281,
334-334, 423-423, and 451-451.

Source: Linters/SAST tools

docs/taos-agent-manual.md-45-46 (1)

45-46: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win

Use ATX headings or make the separators unambiguous.

The markdownlint result reports setext headings in these changed sections. Convert intended headings to ## or ###, or insert a blank line when --- is only a horizontal rule. This removes MD003 and keeps the manual structure stable.

The finding is based on the supplied markdownlint result.

Also applies to: 265-268

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@docs/taos-agent-manual.md` around lines 45 - 46, Update the affected sections
in taos-agent-manual.md to replace intended setext headings with explicit ATX
headings using the appropriate ## or ### level; where --- is intended only as a
horizontal rule, separate it from surrounding content with a blank line.
Preserve the manual’s existing heading hierarchy and content.

Source: Linters/SAST tools

tests/test_model_manifest_integrity.py-200-205 (1)

200-205: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win

Parse download_url as a URL.

startswith("https://") accepts malformed values such as https:// and does not validate a hostname. Parse the value and require an HTTPS scheme plus a non-empty hostname.

This finding uses the supplied URL validation rule.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@tests/test_model_manifest_integrity.py` around lines 200 - 205, Update the
download_url validation rule in the manifest integrity test to parse the value
as a URL rather than using startswith("https://"). Require both an HTTPS scheme
and a non-empty hostname, while preserving the existing error reporting for
invalid URLs.

tests/test_model_manifest_integrity.py-206-210 (1)

206-210: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win

Reject booleans as size_mb.

In Python, bool is a subclass of int. A manifest value of size_mb: true therefore passes the current check and is treated as 1. Use type(size_mb) is int, or explicitly exclude booleans before checking that the value is positive.

This finding uses the supplied size validation expression.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@tests/test_model_manifest_integrity.py` around lines 206 - 210, Update the
size_mb validation in the manifest integrity check to reject boolean values
explicitly, since bool is an int subclass in Python. Use an exact-int check or
exclude booleans before validating that size_mb is positive, while preserving
the existing error reporting.

tinyagentos/registry.py-50-50 (1)

50-50: 🗄️ Data Integrity & Integration | 🟡 Minor | ⚡ Quick win

Validate context_window at the manifest boundary.

The int annotation does not validate YAML data. A manifest can provide null, a string, a negative integer, or a boolean. Such a value violates the 0 = unknown contract and can break context budgeting. Reject or normalize non-negative integers, excluding booleans, before constructing AppManifest.

This finding uses the supplied field declaration and from_dict() assignment.

Also applies to: 79-79

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@tinyagentos/registry.py` at line 50, Validate context_window in from_dict()
before constructing AppManifest: accept only non-boolean integers greater than
or equal to zero, and reject or normalize null, strings, booleans, and negative
values according to the manifest’s established validation behavior. Ensure both
context_window declarations preserve the 0 = unknown contract.

🧹 Nitpick comments (9)

scripts/check_secret_ignores.py (3)

187-197: 🩺 Stability & Availability | 🔵 Trivial | 💤 Low value

Validate --repo-root before running Git.

If the caller passes a path that does not exist, _read_gitignore returns an empty string, and the first _run_git call raises FileNotFoundError or NotADirectoryError from cwd. The user sees a traceback instead of a clear message.

🛡️ Proposed fix

     args = parser.parse_args(argv)
     repo_root = Path(args.repo_root)
+    if not repo_root.is_dir():
+        print(f"secret-ignores-guard: --repo-root is not a directory: {repo_root}")
+        return 2
 
     violations = check_secret_ignores(repo_root)

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@scripts/check_secret_ignores.py` around lines 187 - 197, Validate the parsed
repo_root in main before calling check_secret_ignores or any Git operation:
require it to exist and be a directory, and report an argparse-style clear error
for invalid paths instead of allowing cwd failures. Preserve the existing
behavior for valid repository roots.

---

109-115: 📐 Maintainability & Code Quality | 🔵 Trivial | 💤 Low value

Static analysis flags both Git helpers identically, and both hits are false positives. Each script runs Git through a fixed argument list with no shell and no request-derived input, so S603, S607, and the ast-grep subprocess-from-request rule do not describe a real risk. Add scoped suppressions only if CI enforces these rules.

• scripts/check_secret_ignores.py#L109-L115: add # noqa: S603, S607 to the subprocess.run call in _run_git, or configure a per-file ignore for scripts/.
• scripts/check_store_wiring.py#L47-L48: apply the same suppression to the subprocess.run call in _run_git, and to the git show call at lines 72-75.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@scripts/check_secret_ignores.py` around lines 109 - 115, Add scoped
static-analysis suppressions for the fixed, non-shell Git invocations: annotate
the subprocess.run call in scripts/check_secret_ignores.py (lines 109-115) and
the _run_git subprocess.run call plus git show invocation in
scripts/check_store_wiring.py (lines 47-48 and 72-75) with the required S603 and
S607 ignores, or configure an equivalent per-file ignore for scripts/.

Source: Linters/SAST tools

---

151-154: 🎯 Functional Correctness | 🔵 Trivial | 💤 Low value

Add the -- separator before the path argument.

git check-ignore --quiet <path> treats an argument that starts with - as an option. The current SECRET_PATHS entries are all safe, but a future entry such as -config.key would make Git report an unknown-option error. The script then returns exit code 129, and is_path_ignored reports False. The gate fails with a misleading message instead of a clear error.

🛡️ Proposed fix

 def is_path_ignored(path: str, repo_root: Path) -> bool:
     """True if `git check-ignore` reports `path` as ignored on `repo_root`."""
-    result = _run_git(["check-ignore", "--quiet", path], repo_root)
+    result = _run_git(["check-ignore", "--quiet", "--", path], repo_root)
     return result.returncode == 0

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@scripts/check_secret_ignores.py` around lines 151 - 154, Update
is_path_ignored to pass the Git option terminator "--" immediately before path
in the check-ignore arguments, ensuring paths beginning with "-" are treated as
operands while preserving the existing return-code handling.

docs/RELEASING.md (1)

37-45: 📐 Maintainability & Code Quality | 🔵 Trivial | 💤 Low value

Consider listing the pattern families instead of a partial pattern list.

The prose lists a subset of REQUIRED_PATTERNS. The script also requires *.token, *.cred, data/.secrets_key, data/.seeded-agent-tokens.json, data/secrets.db*, data/*.key, data/*.token, and the new data/.litellm_master_key. A partial list drifts as patterns are added.

📝 Proposed wording

-The `secret-ignores-gate` runs on the `master` push (and on the PR merge result)
-and confirms the promoted `.gitignore` still ignores every secret-shaped path it
-did on `dev` -- `identity.json`, `*.key`, `*.p8`, `*credentials.json`, `*creds*.json`
-and the `*_private.*` key shapes, plus the `secrets/` and `data/hub/` rules. Re-run
-it by hand if a conflict resolution touched `.gitignore`:
+The `secret-ignores-gate` runs on the `master` push (and on the PR merge result)
+and confirms the promoted `.gitignore` still ignores every secret-shaped path it
+did on `dev`. The authoritative list is `REQUIRED_PATTERNS` in
+`scripts/check_secret_ignores.py`; it covers key material (`*.key`, `*.p8`,
+`*_private.*`), credential files (`*credentials.json`, `*creds*.json`,
+`identity.json`, `*.token`, `*.cred`) and the `secrets/`, `data/hub/` and
+`data/` rules. Re-run it by hand if a conflict resolution touched `.gitignore`:

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@docs/RELEASING.md` around lines 37 - 45, Update the secret-ignores-gate
documentation to describe the required pattern families comprehensively rather
than enumerating a partial, hardcoded list. Reference the REQUIRED_PATTERNS
categories, including secret file extensions, credential/key/token paths,
private-key shapes, and secrets/data directory rules, while preserving the
command for rerunning scripts/check_secret_ignores.py.

scripts/check_store_wiring.py (2)

154-162: 🎯 Functional Correctness | 🔵 Trivial | ⚡ Quick win

The name-level fallback lets an unwired store pass the gate.

_is_wired_in_app_py returns True when the class name appears anywhere in tinyagentos/app.py. A bare from tinyagentos.projects.strike_store import StrikeStore satisfies the regex. The store is then never assigned to app.state, and the gate reports it as wired. This is the exact failure mode the guard exists to catch.

Two further weaknesses:

• re.sub(r"#.*", "", app_py_content) also strips # characters inside string literals, so the comment stripping is not reliable.
• A class name mentioned only in a type annotation or a docstring also passes.

Consider reporting the fallback as a warning instead of a pass, or restricting the fallback to lines that also contain app.state.

♻️ Proposed tightening

 def _is_wired_in_app_py(app_py_content: str, class_name: str) -> tuple[bool, str]:
     ast_ok = _is_wired_ast(app_py_content, class_name)
     if ast_ok:
         return True, "AST"
     code_only = re.sub(r"#.*", "", app_py_content)
-    name_ok = bool(re.search(rf"\b{re.escape(class_name)}\b", code_only))
+    # Require the name to co-occur with an app.state assignment on the same
+    # line; a bare import must not satisfy the guard.
+    name_ok = bool(re.search(
+        rf"^.*app\.state\b.*\b{re.escape(class_name)}\b.*$",
+        code_only,
+        re.MULTILINE,
+    ))
     if name_ok:
         return True, "name-level-fallback"
     return False, "unwired"

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@scripts/check_store_wiring.py` around lines 154 - 162, Tighten
_is_wired_in_app_py so a bare class-name match cannot report a store as wired.
Restrict any fallback to reliable, non-comment source that also demonstrates
assignment to app.state, preferably using AST parsing rather than regex comment
stripping; otherwise return unwired and preserve the existing AST success path.

---

165-184: 🎯 Functional Correctness | 🔵 Trivial | 💤 Low value

build_class_hierarchy keys classes by bare name across the whole package.

classes[node.name] = bases collapses same-named classes from different modules. It also resolves a base class by bare name only. A local class Base(...) in one module can make an unrelated class Foo(Base) in another module look like a BaseStore subclass, or hide a real one. The result is a false positive or a false negative in the gate.

For a repository guard this is likely acceptable. If collisions appear, key the map by module:ClassName and resolve bases through the file's imports.

Also applies to: 205-221

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@scripts/check_store_wiring.py` around lines 165 - 184, Update
build_class_hierarchy to identify classes by their module-qualified names rather
than bare class names, preserving separate entries for same-named classes in
different files. Resolve each class’s base references within its module and
through that module’s imports, so local or unrelated classes cannot be mistaken
for the intended store hierarchy; apply the same qualified-name handling to the
downstream logic around the additionally affected range.

tinyagentos/projects/lists_store.py (1)

177-179: 📐 Maintainability & Code Quality | 🔵 Trivial | 💤 Low value

Add strict= to zip() as Ruff B905 requires.

cur.description and row always have equal length here, so strict=True is safe and silences the lint warning.

♻️ Proposed change

-            return dict(zip(keys, row))
+            return dict(zip(keys, row, strict=True))

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@tinyagentos/projects/lists_store.py` around lines 177 - 179, Update the dict
construction in the cursor row conversion to call zip with strict=True,
preserving the existing keys and row pairing while satisfying Ruff B905.

Source: Linters/SAST tools

tests/test_chat_context_window.py (1)

47-55: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win

The suffix assertion is vacuous because all messages are identical.

Every message has the same content, so line 55 passes for any kept subset, including a non-contiguous or oldest-first one. Give each message a distinct marker to prove the kept set is the newest contiguous suffix.

💚 Proposed test change

-    message = "x" * 400  # 100 tokens per message
-    msgs = [_msg("user", message) for _ in range(20)]  # 2000 tokens total
+    # Distinct prefix per message, same token cost (100 tokens each).
+    contents = [f"{i:02d}" + "x" * 398 for i in range(20)]  # 2000 tokens total
+    msgs = [_msg("user", c) for c in contents]
     ctx = build_context_window(msgs, limit=20, max_tokens=budget)
 
     total = sum(estimate_tokens(m["content"]) for m in ctx)
     assert total <= budget
     assert len(ctx) < 20
     # Oldest dropped first -> the kept set is a contiguous suffix.
-    assert [m["content"] for m in ctx] == [message] * len(ctx)
+    assert [m["content"] for m in ctx] == contents[-len(ctx):]

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@tests/test_chat_context_window.py` around lines 47 - 55, Update the test data
in the context-window test around build_context_window so each message has a
distinct, position-dependent marker while retaining the intended token budget.
Change the final assertion to compare the retained messages against the expected
newest contiguous suffix, ensuring the test detects non-contiguous or
oldest-first selection.

tests/test_model_manifest_integrity.py (1)

18-37: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win

Expose catalog targets as data instead of parsing source text.

Changing hardware_to_targets() from literal targets.append(...) calls to a constant, helper, or extend() call can make this test fail or omit valid targets. Expose a shared target enum or constant from tinyagentos/cluster/capabilities.py and import it here. Test hardware_to_targets() behavior separately.

This finding uses the supplied target extraction and capability implementation.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@tests/test_model_manifest_integrity.py` around lines 18 - 37, Replace the
source-text regex extraction used to build KNOWN_TARGETS with an imported shared
target enum or constant exposed by capabilities.py. Update hardware_to_targets()
to derive its accepted targets from that shared definition, and keep this test
focused on manifest integrity while covering hardware_to_targets() behavior
separately.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@docs/design/agent-loop-subagents.md`:
- Around line 48-53: Revise the cancellation design around cancel_subagents()
and reach_safe_point so subagents do not resume the main loop while inside
side-effecting tool or file operations. Define cooperative cancellation or
rollback at those boundaries, and specify the bounded-timeout outcome, including
how unfinished work is handled, before transitioning the loop to IDLE.
- Around line 88-105: Update the desktop endpoint flow around AgentLoop and
reach_safe_point() so every queued message is either redriven as its own agent
turn before the stream closes or explicitly returned as a retry result; do not
leave queued requests handled only by the turn-holder’s stream. Revise the
“Still not integrated” documentation to describe the implemented behavior
accurately and remove the claim that queued messages are merely surfaced without
redrive.

In `@docs/design/radio-studio.md`:
- Around line 64-72: Add normalized vendorid and productid fields to the SdrInfo
dataclass, populated by SDR detection and preserved through the LXC and Docker
passthrough paths. Ensure both backends consume these fields when constructing
USB device descriptors, including the documented empty or unknown-value
behavior.
- Around line 184-202: Update the hackrf_spectrum_survey tool_schema and its
route contract to validate device-derived frequency bounds, require start_hz <
stop_hz, constrain gain and bin_width_hz, and cap the frequency span or
resulting bin count. Ensure the route applies a timeout and propagates
cancellation before forwarding requests, preserving only bounded, valid survey
operations.
- Around line 95-113: Update the container passthrough design around
create_container() and usb_devices so raw USB access cannot bypass receive-only
restrictions. Route SDR access through a host-side receive-only broker or
allowlisted device API, or enforce TX disable in hardware, and document the
selected mechanism. Add a container-level test that attempts a transmit
operation and verifies it is denied.

In `@docs/mirror-policy.md`:
- Around line 25-38: Update the Hailo `.hef` exception documentation to state
that installation is unavailable if the vendor CDN disappears unless a legally
approved fallback is established. Keep the existing SHA-256 integrity checks,
but remove claims that a pinned digest proves licensing or provenance, and avoid
promising future mirroring without an approved source.

In `@tests/test_lxc_installer.py`:
- Around line 674-721: Update the _run_generic_install test helper to patch
containers.destroy_container alongside the other container operations, using an
async mock that prevents real Incus calls during install rollback while
preserving the helper’s existing result and proxy assertions.

In `@tests/test_model_manifest_integrity.py`:
- Line 46: Harden the checksum exemption around _SHA256_ALLOWLIST so newly added
manifest IDs cannot silently bypass sha256 validation. Compare the allowlist
against a reviewed baseline checked by CI, or add an explicit approval mechanism
that fails when the set grows; preserve exemptions only for pre-approved
existing IDs in the checksum validation loop.

In `@tinyagentos/chat/chat_exporter.py`:
- Around line 209-246: Update the oversize-envelope truncation logic in the
exporter method containing the shown block so the budget accounts for JSON
escaping in the final envelope. After constructing the reference note and
candidate body, repeatedly measure _serialized_size(envelope) with the actual
body inserted and shrink the UTF-8 body until the serialized envelope is at or
below self._max_message_bytes, preserving valid UTF-8 and the truncation suffix.

In `@tinyagentos/projects/task_store.py`:
- Around line 475-482: Update the unquarantine flow around the task state commit
and _strikes.clear_strikes so strikes recorded after the task becomes open are
preserved: either clear only strikes created before the transition or perform
the reset atomically before committing/exposing the task to the ready pool. Add
a regression test that interleaves a new verification failure between the state
transition and strike reset and verifies the new strike remains.
- Around line 470-473: Update the unquarantine SQL transition in the
project-task restoration logic to set claimed_by and claimed_at to NULL
alongside status = 'open'. Add a regression test covering the claimed →
quarantined → unquarantined → claimed flow and verify the task can be claimed
again.

In `@tinyagentos/routes/auth.py`:
- Around line 663-664: Reject requests with a missing User-Agent when validating
sessions bound to a user-agent hash: update tinyagentos/routes/auth.py lines
663-664 to enforce this before accepting the session, and apply the same
rejection before accepting the WebSocket in tinyagentos/routes/canvas.py lines
82-88. Add no-User-Agent coverage for /auth/status and /auth/me in
tests/test_auth.py lines 1225-1263.

In `@tinyagentos/routes/channel_hub.py`:
- Around line 268-274: Update validate_session in tinyagentos/auth.py to reject
missing user_agent whenever user_agent_hash exists, while preserving access for
legacy sessions without a stored hash. In tinyagentos/routes/channel_hub.py
lines 268-274 and tinyagentos/routes/chat.py lines 114-120, pass
websocket.headers.get("user-agent") directly without an empty-string fallback so
missing headers remain detectable.

In `@tinyagentos/routes/models.py`:
- Line 44: Update delete_model() to use the same recursive model-discovery and
ownership logic as get_downloaded_models(), including nested .hef files, instead
of a top-level models_dir glob. Ensure successful deletion of a nested .hef file
calls mark_uninstalled(). Add a regression test covering deletion from the
documented nested layout.

In `@tinyagentos/routes/terminal.py`:
- Around line 26-30: Update the WebSocket authentication flow around
auth_mgr.validate_session and the corresponding API middleware to reject
requests with a missing User-Agent when the session is User-Agent-bound, while
continuing to accept legacy sessions without user_agent_hash. Preserve
validate_session behavior for session_user() callers unless they are updated to
pass the request User-Agent, and add regression tests covering bound-session
rejection and legacy-session acceptance.

---

Minor comments:
In @.claude/skills/taos-development-skill/SKILL.md:
- Around line 451-453: Update the fenced block containing the
Store-Unwired-Intentionally example to specify the text language identifier,
using the existing block content unchanged.

In `@CHANGELOG.md`:
- Around line 118-122: Update the changelog entry describing the grouped jsdom
30 Dependabot failure to use past tense for the former Node 20 CI configuration,
while preserving the details about jsdom’s Node 22.13 requirement and the
spa-build context.

In `@desktop/src/stores/__tests__/wallpaper-fit.test.ts`:
- Around line 62-69: Update the “loads the persisted value as the initial store
state” test to verify the Zustand initializer rather than calling
loadWallpaperFit directly: prepopulate localStorage, reset modules with
vi.resetModules(), dynamically import theme-store, and assert
useThemeStore.getState().wallpaperFit equals the persisted value.

In `@desktop/src/stores/theme-store.ts`:
- Around line 196-203: Update getDeviceId() to generate new identifiers with the
existing randomId() helper instead of crypto.randomUUID(), preserving
localStorage persistence and returning the stored identifier. Add a regression
test covering environments where crypto.randomUUID is unavailable.

In `@desktop/src/theme/tokens.css`:
- Around line 206-223: Update the non-tile selectors in the wallpaper fit
rules—fill, fit, stretch, and center—to explicitly set background-repeat to
no-repeat, while preserving repeat only for the tile selector.

In `@docs/design/agent-loop-subagents.md`:
- Line 64: Update all listed Markdown fences to include the required language
identifiers: use a text fence at docs/design/agent-loop-subagents.md:64-64 and
remove the standalone text marker from its block body; use bash fences at
docs/design/radio-studio.md:100-100 and 106-106; and use text fences at
docs/design/radio-studio.md:265-265, 281-281, 334-334, 423-423, and 451-451.

In `@docs/design/radio-studio.md`:
- Around line 325-328: Update the “Agent Chat” command list in the Phase 1
contract to remove or explicitly mark `/stare` and `/capture` as unavailable,
while retaining the supported `/survey` and `/alert` commands.

In `@docs/taos-agent-manual.md`:
- Around line 45-46: Update the affected sections in taos-agent-manual.md to
replace intended setext headings with explicit ATX headings using the
appropriate ## or ### level; where --- is intended only as a horizontal rule,
separate it from surrounding content with a blank line. Preserve the manual’s
existing heading hierarchy and content.

In `@tests/conftest.py`:
- Around line 359-362: Update the client fixture teardown to close task_strikes
after project_task_store.close() by awaiting task_strikes.close(), ensuring the
initialized SQLite connection is released before teardown completes.

In `@tests/test_chat_exporter.py`:
- Around line 227-239: Update the test loop around store.send_message to capture
the returned message identifier, then use that id in the chat_messages UPDATE
instead of selecting the latest row by created_at. Keep assigning ts and
committing each inserted message so the test reliably creates the intended
equal-timestamp scenario.

In `@tests/test_model_manifest_integrity.py`:
- Around line 200-205: Update the download_url validation rule in the manifest
integrity test to parse the value as a URL rather than using
startswith("https://"). Require both an HTTPS scheme and a non-empty hostname,
while preserving the existing error reporting for invalid URLs.
- Around line 206-210: Update the size_mb validation in the manifest integrity
check to reject boolean values explicitly, since bool is an int subclass in
Python. Use an exact-int check or exclude booleans before validating that
size_mb is positive, while preserving the existing error reporting.

In `@tests/test_routes_mcp.py`:
- Around line 43-44: Rename the unused app variable to _app in both
test_list_servers_empty and the corresponding test around lines 91–92, while
preserving the existing app_client unpacking and test behavior.

In `@tinyagentos/agent_loop.py`:
- Line 436: Rename the unused done result to _done in both await_subagent and
await_all_subagents when unpacking asyncio.wait results. Apply the change at
tinyagentos/agent_loop.py lines 436-436 and 481-481; leave pending handling
unchanged.

In `@tinyagentos/chat/chat_exporter.py`:
- Around line 188-191: Update the message-body handling around flatten_body so
whitespace-only flattened text does not trigger ChatExportError or abort
export_channel. Treat body as unusable when it contains only whitespace, then
fall back to the existing content value or an appropriate placeholder before the
content_blocks empty-body guard; preserve the error only when no usable body or
fallback exists.

In `@tinyagentos/projects/strike_store.py`:
- Around line 21-23: Update the _row function’s dict(zip(...)) call to pass
strict=True, ensuring the six expected keys must match exactly six selected row
values.

In `@tinyagentos/registry.py`:
- Line 50: Validate context_window in from_dict() before constructing
AppManifest: accept only non-boolean integers greater than or equal to zero, and
reject or normalize null, strings, booleans, and negative values according to
the manifest’s established validation behavior. Ensure both context_window
declarations preserve the 0 = unknown contract.

---

Nitpick comments:
In `@docs/RELEASING.md`:
- Around line 37-45: Update the secret-ignores-gate documentation to describe
the required pattern families comprehensively rather than enumerating a partial,
hardcoded list. Reference the REQUIRED_PATTERNS categories, including secret
file extensions, credential/key/token paths, private-key shapes, and
secrets/data directory rules, while preserving the command for rerunning
scripts/check_secret_ignores.py.

In `@scripts/check_secret_ignores.py`:
- Around line 187-197: Validate the parsed repo_root in main before calling
check_secret_ignores or any Git operation: require it to exist and be a
directory, and report an argparse-style clear error for invalid paths instead of
allowing cwd failures. Preserve the existing behavior for valid repository
roots.
- Around line 109-115: Add scoped static-analysis suppressions for the fixed,
non-shell Git invocations: annotate the subprocess.run call in
scripts/check_secret_ignores.py (lines 109-115) and the _run_git subprocess.run
call plus git show invocation in scripts/check_store_wiring.py (lines 47-48 and
72-75) with the required S603 and S607 ignores, or configure an equivalent
per-file ignore for scripts/.
- Around line 151-154: Update is_path_ignored to pass the Git option terminator
"--" immediately before path in the check-ignore arguments, ensuring paths
beginning with "-" are treated as operands while preserving the existing
return-code handling.

In `@scripts/check_store_wiring.py`:
- Around line 154-162: Tighten _is_wired_in_app_py so a bare class-name match
cannot report a store as wired. Restrict any fallback to reliable, non-comment
source that also demonstrates assignment to app.state, preferably using AST
parsing rather than regex comment stripping; otherwise return unwired and
preserve the existing AST success path.
- Around line 165-184: Update build_class_hierarchy to identify classes by their
module-qualified names rather than bare class names, preserving separate entries
for same-named classes in different files. Resolve each class’s base references
within its module and through that module’s imports, so local or unrelated
classes cannot be mistaken for the intended store hierarchy; apply the same
qualified-name handling to the downstream logic around the additionally affected
range.

In `@tests/test_chat_context_window.py`:
- Around line 47-55: Update the test data in the context-window test around
build_context_window so each message has a distinct, position-dependent marker
while retaining the intended token budget. Change the final assertion to compare
the retained messages against the expected newest contiguous suffix, ensuring
the test detects non-contiguous or oldest-first selection.

In `@tests/test_model_manifest_integrity.py`:
- Around line 18-37: Replace the source-text regex extraction used to build
KNOWN_TARGETS with an imported shared target enum or constant exposed by
capabilities.py. Update hardware_to_targets() to derive its accepted targets
from that shared definition, and keep this test focused on manifest integrity
while covering hardware_to_targets() behavior separately.

In `@tinyagentos/projects/lists_store.py`:
- Around line 177-179: Update the dict construction in the cursor row conversion
to call zip with strict=True, preserving the existing keys and row pairing while
satisfying Ruff B905.

🪄 Autofix

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 116378a0-c105-4f8c-a440-d4b2c397ab95

📥 Commits

Reviewing files that changed from the base of the PR and between f78ff3a and 9eaefde.

⛔ Files ignored due to path filters (2)

• desktop/package-lock.json is excluded by !**/package-lock.json, !**/package-lock.json
• uv.lock is excluded by !**/*.lock, !**/uv.lock

📒 Files selected for processing (82)

• .claude/skills/taos-development-skill/SKILL.md
• .github/workflows/ci.yml
• .github/workflows/secret-ignores-gate.yml
• .github/workflows/store-wiring-gate.yml
• .gitignore
• CHANGELOG.md
• README.md
• app-catalog/models/deepseek-r1-distill-qwen-1.5b-hef/manifest.yaml
• app-catalog/models/llama-3.2-3b-instruct-hef/manifest.yaml
• app-catalog/models/qwen2-1.5b-instruct-hef/manifest.yaml
• app-catalog/models/qwen2.5-1.5b-instruct-hef/manifest.yaml
• app-catalog/models/qwen2.5-coder-1.5b-instruct-hef/manifest.yaml
• desktop/package.json
• desktop/src/App.tsx
• desktop/src/apps/GuidesApp.test.tsx
• desktop/src/apps/SettingsApp.tsx
• desktop/src/apps/agents/RegistryPanel.test.tsx
• desktop/src/apps/agents/RegistryPanel.tsx
• desktop/src/components/Desktop.tsx
• desktop/src/stores/__tests__/wallpaper-fit.test.ts
• desktop/src/stores/theme-store.ts
• desktop/src/theme/tokens.css
• docs/RELEASING.md
• docs/agent-coordination.md
• docs/agent-manual/01-rules.md
• docs/agent-manual/10-image-prompting.md
• docs/agent-manual/index.md
• docs/design/agent-loop-subagents.md
• docs/design/radio-studio.md
• docs/mirror-policy.md
• docs/taos-agent-manual.md
• pyproject.toml
• scripts/check_secret_ignores.py
• scripts/check_store_wiring.py
• tests/catalog/test_resolver_hailo.py
• tests/conftest.py
• tests/projects/test_lists_store.py
• tests/projects/test_strike_wiring.py
• tests/test_agent_chat_router.py
• tests/test_agent_loop.py
• tests/test_agent_registry.py
• tests/test_agent_scope_requests.py
• tests/test_auth.py
• tests/test_chat_context_window.py
• tests/test_chat_exporter.py
• tests/test_check_secret_ignores.py
• tests/test_check_store_wiring.py
• tests/test_lxc_installer.py
• tests/test_model_manifest_integrity.py
• tests/test_registry.py
• tests/test_registry_governance_lifecycle.py
• tests/test_routes_agent_model_api.py
• tests/test_routes_agent_org.py
• tests/test_routes_mcp.py
• tests/test_routes_memory.py
• tests/test_routes_project_invites.py
• tests/test_taos_agent_chat.py
• tests/test_task_store.py
• tests/test_token_rotation.py
• tinyagentos/__init__.py
• tinyagentos/agent_chat_router.py
• tinyagentos/agent_loop.py
• tinyagentos/app.py
• tinyagentos/auth_middleware.py
• tinyagentos/chat/chat_exporter.py
• tinyagentos/cluster/capabilities.py
• tinyagentos/projects/ids.py
• tinyagentos/projects/lists_store.py
• tinyagentos/projects/strike_store.py
• tinyagentos/projects/task_store.py
• tinyagentos/registry.py
• tinyagentos/routes/agent_auth_requests.py
• tinyagentos/routes/agent_registry.py
• tinyagentos/routes/auth.py
• tinyagentos/routes/canvas.py
• tinyagentos/routes/channel_hub.py
• tinyagentos/routes/chat.py
• tinyagentos/routes/memory.py
• tinyagentos/routes/models.py
• tinyagentos/routes/projects.py
• tinyagentos/routes/taos_agent.py
• tinyagentos/routes/terminal.py

[jaylfc]

Adjudication of the 20 bot threads before resolving them (this promotion PR re-reviews the cumulative diff of 27 already-merged dev PRs; each finding was judged against the original change's review, not dismissed on venue):

Real and carded — tsk-y6x6s4 (lead-held, red-first): the unquarantine_task cluster (CodeRabbit x2, Kilo x2 on task_store.py). Verified from source: unquarantine never clears claimed_by, claim_task requires claimed_by IS NULL, so a claimed-then-quarantined card returns "open" but is permanently unclaimable and invisible to the ready pool. The clear_strikes-after-commit window and the hardcoded audit from_status ride in the same card.

Deliberate design, already settled on the original PRs:

• The five "404 breaking change" threads (approve/deny scope-request, revoke, rotate-tokens, org update): existence-hiding is the point of Existence-hiding 404 across ALL agent-registry routes (issue #2106, reported by hognek) #2356 (issue feat(agents): consistent existence-hiding 404 across all agent-registry routes #2106, reported externally). The 403-vs-404 distinction was the vulnerability, not a feature; tests document the new contract byte-for-byte.
• except BaseException in lists_store reorder: deliberately widened in the fix(lists): get_entry cursor scope + reorder rollback on failure (tsk-u23vjy) #2361 lead re-cut so cancellation/exit still rolls back before re-raising; the handler re-raises, nothing is swallowed.
• except HTTPException in agent_auth_requests: a provider outage is an infrastructure failure, not an auth outcome — a 500 there is correct and existence-uniform (it fires identically for existing and nonexistent ids), so the 404 invariant is not breached.

Spec commentary on design docs (radio-studio.md, agent-loop-subagents.md, mirror-policy.md): these documents are pre-build specs with their own cards; the concrete points (survey input bounds, SdrInfo vendor/product ids, subagent cancellation safe-points, queued-message redrive) are captured as build-time acceptance criteria on those tracks, not code defects in this tree.

Test-quality notes (lxc_installer patch-set comparison, sha256 allowlist baseline, chat_exporter escaping budget): noted; the exporter truncation-vs-escaping point is plausible but bounded (envelope overshoot by escape bytes), queued for the next test-debt pass rather than blocking a version promotion.

Resolving all threads on this basis; the release content itself was reviewed PR-by-PR on dev.