feat(miner): run the target repo's own test/lint/build commands before opening a PR#8819
Merged
Conversation
…e opening a PR (#8807) Nothing independently verified a coding agent's work before submission: the only verification module (engine lint-guard) is hardcoded to loopover's own monorepo commands and never passed in production, and coding-task-spec's validation guidance only TELLS the agent which commands to run. A known-bad change passed every AMS-side gate on the agent's self-attestation alone. - target-repo-verification.ts: runs stack-detection's already-inferred commands (test → lint → build, highest signal first) from the attempt's worktree, per-command timeout (10 min), stop at first failure, bounded output tail. An undetected stack or empty command set SKIPS (recorded, never a failure) — the gate is only as smart as detection, and empty detection must not block repos with unconventional tooling. - attempt-runner: the gate runs after handoff + kill-switch recheck and BEFORE the freshness read (a failing build never spends GitHub budget); a failure returns the new verification_failed outcome — the attempt never submits, the worktree is retained for postmortem, and the existing not-submitted notification plumbing carries the reason. Deliberately not re-entering the iterate loop in this change (the loop's own self-review iterations already ran; never-submit-known-bad is the trust win) — loop feedback is the tracked follow-up on the issue. - attempt-cli binds the worktree-scoped thunk (same stack detection the agent's guidance rendered), with MINER_SKIP_TARGET_REPO_VERIFICATION as the escape hatch for suites exceeding the per-command bound.
Contributor
|
Superagent didn't find any vulnerabilities or security issues in this PR. |
Deploying with
|
| Status | Name | Latest Commit | Preview URL | Updated (UTC) |
|---|---|---|---|---|
| ✅ Deployment successful! View logs |
loopover-ui | 7127ce4 | Commit Preview URL Branch Preview URL |
Jul 26 2026, 08:28 AM |
Bundle ReportChanges will increase total bundle size by 72 bytes (0.0%) ⬆️. This is within the configured threshold ✅ Detailed changes
Affected Assets, Files, and Routes:view changes for bundle: loopover-uiAssets Changed:
|
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #8819 +/- ##
==========================================
+ Coverage 90.56% 92.97% +2.40%
==========================================
Files 96 704 +608
Lines 22490 57313 +34823
Branches 3884 20306 +16422
==========================================
+ Hits 20367 53284 +32917
- Misses 1945 3147 +1202
- Partials 178 882 +704
Flags with carried forward coverage won't be shown. Click here to find out more.
|
loopover-orb
Bot
added
the
gittensor:bug
Contributor
|
Warning ⏸️ LoopOver review result - manual review recommendedReview updated: 2026-07-26 09:03:13 UTC
Review summary Nits — 5 non-blocking
Decision drivers
Context & advisory signals — never blocks the verdict
Linked issue satisfactionPartially addressed Review context
Contributor next steps
Signal definitions
🧪 Chat with LoopOverAsk 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.
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.
|
…settle bounded
The default verification spawn killed only the shell at the timeout; on
Linux the command's descendants survive, keep the stdio pipes open, and
stall the close event (observed as a 30s hang in CI). Spawn detached,
kill the negative pid, and add a bounded post-kill settle so the gate
can never outlive its per-command timeout even if an orphan escapes the
group.
Also fix the queue.test.ts/queue-2.test.ts fetch stubs whose generic {}
fallback broke the close-explanation marker search (the comment list
endpoint must return an array) — a latent mock-shape gap surfaced by
routing enforced closes through createOrUpdateCloseExplanationComment.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
The final round-2 audit item (#8807), and the biggest AMS trust gap: submission quality rested entirely on the coding agent's self-attestation — the only real verification module (engine
lint-guard) is loopover-monorepo-hardcoded and never passed in production, and the task spec's validation guidance only tells the agent what to run.Changes
target-repo-verification.ts(new): runs the target repo's own detected commands (stack-detection's inferred test → lint → build, same source the agent's guidance renders) from the attempt worktree via a shell spawn seam — per-command 10-minute bound, timeout-kill = failure, stop-at-first-failure, 4KB output tail for postmortem. Undetected stack / no inferred commands ⇒ recorded SKIP, never a failure (empty detection must not block unconventional repos).attempt-runner.ts— optionalverifyTargetRepodep, gated after handoff + kill-switch recheck and before the freshness read (a failing build never spends GitHub API budget). Failure ⇒ newverification_failedoutcome: no submission, worktree retained (attemptOk=false path), existing not-submitted notification plumbing carries the reason. Absent dep = byte-identical pre-feat(miner): run the TARGET repo's own test/build commands before opening a PR — stop trusting the agent's self-attestation #8807 flow (pinned).attempt-cli.ts— binds the worktree-scoped thunk;MINER_SKIP_TARGET_REPO_VERIFICATIONescape hatch for suites exceeding the bound; injection seam mirroring the sibling resolvers.Test plan
npm run typecheckclean;@loopover/minerbuildsverification_failedblocks before the freshness read and the PR write; passed/skipped/absent-dep all submit unchangedCloses #8807