fix(commands): authorize intent routing before AI classification

[JSONbored]

Motivation

• Prevent a pre-authorization AI-consumption path introduced by intent routing where unrecognized @gittensory mentions could trigger env.AI_ADVISORY.run before command authorization and per-actor throttling.
• Ensure enabling advisoryAiRouting.intentRouting alone cannot create an unmetered cost-bearing surface when commandRateLimitPolicy is left off.

Description

• Require the commenter to be authorized for at least one INTENT_ROUTABLE_COMMANDS entry before calling the intent classifier by checking isAuthorizedCommandActor over that closed set and only invoking maybeThrottleIntentRouting/classifyGittensoryIntent when authorized.
• Treat intent-routing as throttled when commandRateLimitPolicy is off (make maybeThrottleIntentRouting return true for the off policy) so intent routing does not bypass rate-limit billing.
• Reuse the pre-classification miner-detection (official) result and delay duplicate miner checks so the later real authorization step can reuse the same value rather than re-fetching.
• Add/adjust unit test coverage and fixtures: update test/unit/queue-5.test.ts to include a regression test that an unauthorized commenter does not call AI_ADVISORY.run, enable the hold policy where needed, and extend the mention payload helper to allow custom commenter logins.

Testing

• Ran the new regression test with npx vitest run test/unit/queue-5.test.ts -t "unauthorized commenters cannot spend" --reporter=verbose and it passed.
• Type-checked the code with npx tsc --noEmit --pretty false which succeeded.
• Verified no diff/check issues with git diff --check (no problems reported).

---

Codex Task

[superagent-security]

Superagent didn't find any vulnerabilities or security issues in this PR.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.34%. Comparing base (810b203) to head (b4e9f69).
⚠️ Report is 3 commits behind head on main.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #5107   +/-   ##
=======================================
  Coverage   94.33%   94.34%           
=======================================
  Files         473      473           
  Lines       39930    39936    +6     
  Branches    14559    14562    +3     
=======================================
+ Hits        37669    37676    +7     
  Misses       1585     1585           
+ Partials      676      675    -1     

Flag	Coverage Δ
shard-1	46.33% <30.00%> (-0.01%)	⬇️
shard-2	34.43% <100.00%> (-0.25%)	⬇️
shard-3	31.22% <0.00%> (-0.82%)	⬇️
shard-4	32.66% <0.00%> (+0.71%)	⬆️
shard-5	33.66% <0.00%> (-0.10%)	⬇️
shard-6	45.08% <0.00%> (+0.20%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
src/queue/processors.ts	95.50% <100.00%> (+0.04%)	⬆️

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[gittensory-orb]

Caution

🟥🟥🟥🟥🟥🟥🟥🟥🟥🟥🟥🟥

🛑 Gittensory review result - fixes required

_{Review updated: 2026-07-12 08:35:23 UTC}

2 files · 1 AI reviewer · 2 blockers · readiness 93/100 · CI failing · unstable

🛑 Suggested Action - Manual Review

• No linked issue detected — If this PR is intended to solve an issue, link it explicitly in the PR body.
• Maintainer requires a linked issue — Link the relevant issue (for example Closes #123) before opening the PR.

Review summary
This closes a real pre-authorization AI-spend hole: the intent classifier for unrecognized @​gittensory mentions could previously fire before any per-actor authorization or rate-limit check, and `commandRateLimitPolicy: off` let it bypass throttling entirely. The fix requires the commenter to be authorized for at least one `INTENT_ROUTABLE_COMMANDS` entry (via `isAuthorizedCommandActor`) before calling `maybeThrottleIntentRouting`/`classifyGittensoryIntent`, and flips `maybeThrottleIntentRouting` to return `true` (blocking) when policy is `off`, plus reuses the pre-classification miner-detection result to avoid a duplicate fetch. The logic traces correctly end-to-end and a targeted regression test asserts `AI_ADVISORY.run` is never invoked for an unauthorized commenter, with existing intent-routing tests updated to set `commandRateLimitPolicy: hold` so they still exercise the classifier path.

Blockers

• src/queue/processors.ts:12505: `authorizedForIntentRouting` calls `isAuthorizedCommandActor` without `commandRateLimitPolicy` and `pullRequestOpenAndNotDraft`, so a PR author who is authorized for an intent-routable command that depends on those fields, such as `chat`, is treated as unauthorized before classification and falls back instead of reaching the real authorization path; pass the same fields used by the later authorization call, e.g. `commandRateLimitPolicy: settings.commandRateLimitPolicy` and the computed open/non-draft PR boolean.

Nits — 5 non-blocking

• The new `mentionPayload` still hardcodes `author_association: "OWNER"` regardless of the `commenter` override (test/unit/queue-5.test.ts:1150), so the 'unauthorized commenter' regression test relies on the mocked `/collaborators/.../permission` → "none" response rather than a genuinely non-owner association — worth varying the association with the commenter for clarity even though the live-permission mock makes the test pass correctly.
• codecov/patch is at 80% against a 99% target for this diff; the `needsIntentMinerDetection` branch and the `official` reuse/no-reuse branches in maybeProcessGittensoryMentionCommand (src/queue/processors.ts) look like likely uncovered paths worth adding direct-unit coverage for.
• The flagged possible secret at test/unit/queue-5.test.ts:1410 is almost certainly `generatePrivateKeyPem()` test fixture output, not a real credential, but worth a quick glance to confirm nothing static leaked in.
• Add a unit test where `INTENT_ROUTABLE_COMMANDS.some(...)` returns true for a commenter authorized on a non-default command to confirm the `authorizedForIntentRouting` gate isn't accidentally too narrow.
• Consider a short comment near the new `needsIntentMinerDetection`/`official` prefetch (src/queue/processors.ts:12492) noting it's an optimization only — the later block still safely refetches when `official` wasn't prefetched — to preempt confusion about the two-step computation.

Why this is blocked

• No linked issue detected — If this PR is intended to solve an issue, link it explicitly in the PR body.
• Maintainer requires a linked issue — Link the relevant issue (for example Closes #123) before opening the PR.

CI checks failing

• codecov/patch — 80.00% of diff hit (target 99.00%)

Signal	Result	Evidence
Code review	❌ 2 blockers	1 reviewer
Linked issue	⚠️ Missing	No linked issue or no-issue rationale found.
Related work	✅ No active overlap found	No same-issue or scoped active PR overlap found.
Change scope	✅ 20/20	Low review scope from cached public metadata (no linked issue context).
Validation posture	✅ 25/25	PR body includes validation/test evidence.
Contributor workload	✅ 10/10	Author activity: 45 registered-repo PR(s), 37 merged, 411 issue(s).
Contributor context	✅ Confirmed Gittensor contributor	JSONbored; Gittensor profile; 45 PR(s), 411 issue(s).
Gate result	❌ Blocking	Repo-configured hard blocker found.
Improvement	✅ Minor	risk: clean · value: minor — Code changes are accompanied by test evidence.

Review context

• Author: JSONbored
• Role context: owner (maintainer lane)
• Public audience mode: oss maintainer
• Lane context: Repository is configured for direct PR review.
• Public profile languages: not available
• Official Gittensor activity: 45 PR(s), 411 issue(s).
• PR-specific overlap: none found.

Contributor next steps

• Treat this as maintainer-lane context rather than normal contributor-lane activity.
• Explain no-issue PR.
• Link the issue being solved, or explicitly explain why this is a no-issue PR.

Signal definitions

• Related work = same linked issue, overlapping active PRs, or title/path similarity.
• Change scope = cached public metadata such as size labels, draft state, and review-burden hints.
• Validation posture = whether the PR provides enough public validation/test evidence for maintainer review.
• Contributor workload = public contributor activity and cleanup pressure, not a repo-wide quality failure.
• Contributor context = public GitHub/Gittensor identity context; non-Gittensor status is not a blocker.

[BETA] Chat with Gittensory

Ask Gittensory a question about this PR directly in a comment — grounded only in the same cached, public-safe facts shown above, never a new claim.

• @gittensory ask <question> answers contribution-quality Q&A with source citations and freshness.
• @gittensory chat <question> answers in natural prose from cached decision-pack facts via local inference (maintainer/collaborator; read-only).
• A plain-language @gittensory mention with a real question is routed to the closest matching read-only command automatically -- no exact syntax required.

Full command reference: https://gittensory.aethereal.dev/docs/gittensory-commands

_{🟩 Safe / merged · 🟦 Advisory · 🟨 Held for review · 🟥 Blocked / closed}

---

💰 Earn for open-source contributions like this. Gittensor lets GitHub contributors earn for the work they already do — register to start earning →.

Checked by Gittensory, a quiet PR intelligence layer for OSS maintainers.

• Re-run Gittensory review