feat(mcp): add repo-agnostic AI issue-planning tool

[JSONbored]

Summary

• Adds loopover_plan_repo_issues, a new MCP tool: given a maintainer-supplied free-form planning goal, AI-generates a small set of structured GitHub issue drafts (title/body/labels) for any repo the caller's App/Orb-brokered install is actually installed on.
• This is genuinely generative — unlike generateContributorIssueDrafts (src/services/contributor-issue-draft.ts), which derives candidates purely from loopover-specific static signals (policy readiness, upstream drift, focus-manifest wanted paths) at zero LLM cost, the new src/services/issue-plan-draft.ts calls the configured AI reviewer, mirroring src/review/planner.ts's @loopover plan model-call/retry/budget shape closely (same primary+fallback model, same rate-limit short-circuit, same AI_DAILY_NEURON_BUDGET accounting).
• Dry-run by default; an explicit {create:true, dryRun:false} is required to write, and creation goes exclusively through the installation-token/Orb-broker path added in #7425/#7467 — never a flat PAT — so it only works on a repo the caller's own App/Orb actually covers.
• The MCP response includes each draft's full title/body/labels (unlike generateContributorIssueDrafts's deliberate scrub of that field): the content here is generated fresh from the caller's own goal for their own repo, so there is no loopover-internal signal to hide, and the whole point of the dry-run-by-default posture is letting a maintainer actually read the proposal before deciding to create it.
• Reuses the dedup/decline machinery's shape (marker/fingerprint/wontfix/cooldown) from contributor-issue-draft.ts but with its own marker prefix (loopover-issue-plan-draft) — that module's own marker functions are hardcoded to its own prefix, so directly reusing them would mistag these drafts.
• Deliberate scope decision, documented for future reference: no dedicated per-repo .loopover.yml settings.advisoryAiRouting.issuePlanning-style enable flag was added, unlike the other 4 advisory-tier AI capabilities (slop, e2eTestGen, planner, summaries). This capability is opt-in simply by a maintainer calling the tool (gated by requireRepoManageAccess), and additionally respects the existing fleet-wide AI_SUMMARIES_ENABLED/AI_PUBLIC_COMMENTS_ENABLED kill switches — a reasonable substitute given the primary safety layer here is access control, not automatic/webhook-triggered firing. A dedicated advisory-routing flag (full config-as-code wiring: schema + resolver + .loopover.yml schema + docs) is a good candidate for a fast-follow if a self-hoster asks for cost-conscious local-model routing on this specific capability.
• Sub-issue of #7424 (Epic: ORB self-hoster issue & milestone planning routine). Milestone assignment is out of scope here — tracked separately as #7427.

Scope

• The PR title follows type(scope): short summary Conventional Commit format, for example fix(api): restore profile access checks.
• This PR is focused and does not mix unrelated backend, UI, MCP, docs, dependency, and deploy changes.
• This follows CONTRIBUTING.md and does not reintroduce GitHub Pages, VitePress, site/, or CNAME.
• I linked a currently open issue this PR resolves (e.g. Closes #123) — a linked open issue is required for every contributor PR.

Validation

• git diff --check
• npm run typecheck
• npm run test:coverage locally — full unsharded suite: 1011 files / 19224 tests passed, 2 skipped. The same 9 pre-existing failures in 3 unrelated miner test files (local-branch.test.ts, miner-init-verify-token.test.ts, miner-self-review-context.test.ts) reproduce identically on a clean checkout with this branch's changes stashed out (confirmed separately, tracked independently, unrelated to this PR). New code (src/services/issue-plan-draft.ts, the new MCP wiring in src/mcp/server.ts) is 100% branch-covered.
• npm run selfhost:env-reference:check / npm run ui:openapi:check / npm run docs:drift-check / npm run manifest:drift-check — all clean; no new env vars introduced (reuses the already-documented AI_SUMMARIES_ENABLED/AI_PUBLIC_COMMENTS_ENABLED/AI_DAILY_NEURON_BUDGET/AI_GATEWAY_ID), no OpenAPI/manifest surface touched (MCP-only).
• npm run ui:lint / npm run ui:typecheck / npm run ui:build — not run; this PR touches no files under apps/loopover-ui/**.
• New or changed behavior has unit/integration tests for new branches, fallback paths, and sanitizer boundaries — test/unit/issue-plan-draft.test.ts (29 tests, 100% branch coverage of the new service) and test/unit/mcp-plan-repo-issues.test.ts (7 tests: access control, the explicit_create_requires_dry_run_false guard, goal validation, and surface parity with the underlying service).

If any required check was skipped, explain why:

• ui:lint/ui:typecheck/ui:build: no apps/loopover-ui/** files changed.

Safety

• No secrets, wallet details, hotkeys, coldkeys, user PATs, private keys, raw trust scores, private rankings, or private maintainer evidence are exposed.
• Public GitHub text stays sanitized, low-noise, and does not imply compensation guarantees or optimization tactics — every generated draft's title/body is checked with isFocusManifestPublicSafe before ever being proposed or created, matching contributor-issue-draft.ts's own contract.
• Auth, cookie, CORS, GitHub App, Cloudflare, or session changes include negative-path tests — MCP access-control denial, missing installation, non-2xx GitHub response, and global agent pause/freeze overlay are all covered.
• API/OpenAPI/MCP behavior is updated and tested where needed — new MCP tool fully tested; no HTTP/OpenAPI surface added.
• UI changes use live API data or real empty/error/loading states. (N/A — no UI.)
• Visible UI changes include a UI Evidence section. (N/A — backend/MCP-only change.)
• Public docs/changelogs are updated where needed. (N/A — no user-facing docs surface for MCP tool additions in this repo beyond the tool's own description string, which is included in this diff.)

UI Evidence

N/A — this is a backend/MCP-only change with no visible UI, frontend, or docs surface.

Notes

• Sub-issue of #7424. Builds directly on #7425's installation-token/Orb-broker issue-write path (already merged, #7467).

Closes #7426

[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 91.37%. Comparing base (582fc3a) to head (495c91e).
⚠️ Report is 4 commits behind head on main.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@           Coverage Diff            @@
##             main    #7476    +/-   ##
========================================
  Coverage   91.36%   91.37%            
========================================
  Files         717      718     +1     
  Lines       73016    73178   +162     
  Branches    21631    21678    +47     
========================================
+ Hits        66713    66864   +151     
- Misses       5265     5272     +7     
- Partials     1038     1042     +4     

Flag	Coverage Δ
shard-1	35.26% <93.20%> (+1.10%)	⬆️
shard-2	41.22% <9.87%> (+1.91%)	⬆️
shard-3	37.29% <9.87%> (+4.69%)	⬆️
shard-4	40.94% <9.87%> (+1.41%)	⬆️
shard-5	27.47% <9.87%> (-9.12%)	⬇️
shard-6	29.78% <59.87%> (-4.55%)	⬇️

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

Files with missing lines	Coverage Δ
src/mcp/server.ts	96.45% <100.00%> (+0.05%)	⬆️
src/services/issue-plan-draft.ts	100.00% <100.00%> (ø)

... and 2 files with indirect coverage changes

[loopover-orb]

Tip

✅ LoopOver review result - approve/merge recommended

_{Review updated: 2026-07-20 13:56:55 UTC}

4 files · 1 AI reviewer · no blockers · readiness 98/100 · CI green · clean

✅ Suggested Action - Approve/Merge

• safe to merge

Review summary
Adds a new MCP tool `loopover_plan_repo_issues` that AI-generates GitHub issue drafts from a free-form maintainer goal, for any repo the caller's App/Orb install covers. It closely mirrors the existing `review/planner.ts` model-call/retry/budget shape and `contributor-issue-draft.ts`'s dedup/decline machinery (with its own marker prefix to avoid tag collisions), defaults to dry-run, and gates real writes behind an explicit `{create:true, dryRun:false}` plus the existing installation-token/Orb-broker path (never a flat PAT). The dry-run guard, global agent pause/freeze override, budget check before the model call, and per-candidate malformed-output tolerance are all traced correctly and covered by extensive unit and MCP-layer tests, including a regression test for the rate-limit short-circuit behavior.

Nits — 6 non-blocking

• src/services/issue-plan-draft.ts:179-215 nests to depth 5 in the main draft loop (model call → per-candidate normalize → safety/duplicate/declined/create branches); consider extracting the per-candidate classification into a small helper for readability.
• runIssuePlanModel (src/services/issue-plan-draft.ts) retries the SAME model on a legitimate `{"issues":[]}` response (e.g. an intentionally too-vague goal) exactly like a parse failure, burning an extra call/budget before falling back — worth special-casing an explicit empty-array response as 'no issues' rather than 'retry'.
• recordIssuePlanUsage always logs `model` as the combined 'primary+fallback' string even when only the primary succeeded, so usage telemetry can't tell which model actually produced the output.
• Several new magic numbers (2000/5/429/0.2/200/7425) could be named constants for consistency with the module's existing MAX_* constants.
• src/mcp/server.ts and issue-plan-draft.ts are both large files; the new tool adds a reasonable, self-contained chunk but keep an eye on server.ts's growth as more tools land there.
• PR author also opened the linked issue — Link an issue that was opened by a different contributor, or provide a rationale for why this self-authored issue represents genuine discovery work.

Decision drivers

• ✅ Code review — No blockers (1 reviewer)
• ✅ Gate result — Passing (No configured blocker found.)

Context & advisory signals — never blocks the verdict

Signal	Result	Evidence
Linked issue	✅ Linked	#7426
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 (1 linked issue).
Validation posture	✅ 25/25	PR body includes validation/test evidence.
Contributor workload	✅ 10/10	Author activity: 28 registered-repo PR(s), 21 merged, 427 issue(s).
Contributor context	✅ Confirmed Gittensor contributor	JSONbored; Gittensor profile; 28 PR(s), 427 issue(s).
Improvement	✅ Minor	risk: clean · value: minor

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: Python, TypeScript, Ruby, Go, JavaScript, MDX, Shell, Solidity
• Official Gittensor activity: 28 PR(s), 427 issue(s).
• PR-specific overlap: none found.

Contributor next steps

• Start here: Treat this as maintainer-lane context rather than normal contributor-lane activity.

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.

🧪 Chat with LoopOver

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

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

Full command reference: https://loopover.ai/docs/loopover-commands

🧪 Experimental — new and may change.

_{🟩 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 LoopOver, a quiet PR intelligence layer for OSS maintainers.

• Re-run LoopOver review