CtrlK
BlogDocsLog inGet started
Tessl Logo

uinaf/autoreview

Runs one independent review of completed local changes, pull requests, branch diffs, or commits through the installed autoreview Go CLI using Codex, Claude, Cursor, or Grok; gathers authoritative acceptance criteria, selects or honors one provider, validates findings, applies scoped fixes, verifies and reruns accepted fixes, and safely reports reproducible CLI defects. Use when the user asks for an autoreview, a review of their code, pull request, or changes, an automated PR or second-model code review, a final tool-backed review, or review closeout after builder verification. Do not use as builder verification or a multi-reviewer panel.

Quality

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Low

Low-risk findings worth noting

Overview
Quality
Evals
Security
Files

results.mdreferences/

Results and failures

Terminal and JSON output represent the same locally validated report. Review results go to stdout; progress and diagnostics go to stderr. When review --output json is selected unambiguously, representable argument, flag, and configuration failures also use the canonical failure report instead of prose. Help remains a separate human-readable surface.

ExitStatusMeaning
0cleanValid review with no findings
1findingsValid review with one or more findings
2failureNo trustworthy review result

Every valid finding is reported regardless of confidence. Findings include priority P0 through P3, confidence, category, and a repository-relative location. The local validator rejects paths or line ranges outside the frozen target.

Failure classes are config, target, secret_scan, capability, authentication, timeout, cancelled, provider, protocol, source_changed, and internal. Do not reinterpret an operational failure as a clean review.

Autoreview performs at most one configured retry, only for a malformed protocol response, using the same frozen bundle and provider. Authentication, capability, timeout, cancellation, and provider failures are not retried.

Cursor first requires one canonical review object. Its only recovery accepts plain prose followed by exactly one canonical trailing object. Fences, ambiguous braces, JSON-value prefixes, multiple objects, and suffix prose fail closed. Successful local recovery is recorded as cursor_trailing_object.

After any provider return, autoreview recollects the target. A changed snapshot produces source_changed; discard the findings and rerun from a new freeze.

SKILL.md

tile.json