|
| 1 | +--- |
| 2 | +name: autoreview |
| 3 | +description: "Auto Review closeout. Codex review is the default when no engine is set and is the recommended reviewer." |
| 4 | +--- |
| 5 | + |
| 6 | +# Auto Review |
| 7 | + |
| 8 | +Run the bundled structured review helper as a closeout check. This is code review, not Guardian `auto_review` approval routing. |
| 9 | + |
| 10 | +Codex review is the default when no engine is set. It usually delivers the best review results and should remain the normal final closeout engine. |
| 11 | + |
| 12 | +Use when: |
| 13 | + |
| 14 | +- user asks for Codex review / Claude review / autoreview / second-model review |
| 15 | +- after non-trivial code edits, before final/commit/ship |
| 16 | +- reviewing a local branch or PR branch after fixes |
| 17 | + |
| 18 | +## Contract |
| 19 | + |
| 20 | +- Treat review output as advisory. Never blindly apply it. |
| 21 | +- Verify every finding by reading the real code path and adjacent files. |
| 22 | +- Read dependency docs/source/types when the finding depends on external behavior. |
| 23 | +- Reject unrealistic edge cases, speculative risks, broad rewrites, and fixes that over-complicate the codebase. |
| 24 | +- Prefer small fixes at the right ownership boundary; no refactor unless it clearly improves the bug class. |
| 25 | +- Keep going until structured review returns no accepted/actionable findings. |
| 26 | +- If a review-triggered fix changes code, rerun focused tests and rerun the structured review helper. |
| 27 | +- For security-audit suppression changes, verify accepted findings remain auditable: suppressed findings stay in structured output, active output keeps an unsuppressible suppression notice, and aggregate findings cannot hide unrelated active risk. |
| 28 | +- Never switch or override the requested review engine/model. If the review hits model capacity, retry the same command a few times with the same engine/model. |
| 29 | +- Be patient with large bundles. Structured review can take up to 30 minutes while the model call is active, especially with Codex tools or web search. |
| 30 | +- Treat heartbeat lines like `review still running: ... elapsed=... pid=...` as healthy progress, not a hang. Let the helper continue while heartbeats are advancing. Pass `--stream-engine-output` when live engine text is useful; Codex and Claude filter tool/file chatter, other engines pass raw output through. |
| 31 | +- Do not kill a review just because it has been quiet for 2-5 minutes, or because it is still running under the 30-minute window. Inspect the process only after missing multiple expected heartbeats, after 30 minutes, or after an obviously failed subprocess; prefer letting the same helper command finish. |
| 32 | +- Tools are useful in review mode. The helper allows read-only inspection tools and web search by default so reviewers can check dependency contracts, upstream docs, and current behavior. |
| 33 | +- Security perspective is always included, but it should not cripple legitimate functionality. Report security findings only when the change creates a concrete, actionable risk or removes an important safety check. |
| 34 | +- For regression provenance, if no blamed PR is traceable, use the blamed commit as the provenance: commit SHA, date, and author username. Do not guess a merger or frame missing PR metadata as a separate finding. |
| 35 | +- Do not invoke built-in `codex review`, nested reviewers, or reviewer panels from inside the review. The helper builds one bundle, calls one selected engine, validates one structured result, and stops. |
| 36 | +- Stop as soon as the helper exits 0 with no accepted/actionable findings. Do not run an extra review just to get a nicer "clean" line, a second opinion, or clearer closeout wording. |
| 37 | +- Treat the helper's successful exit plus absence of actionable findings as the clean review result, even if the underlying Codex CLI output is terse. |
| 38 | +- Multi-reviewer panels are opt-in only. Use them when explicitly requested or when risk justifies the extra spend; the main agent still verifies every accepted finding before fixing. |
| 39 | +- If rejecting a finding as intentional/not worth fixing, add a brief inline code comment only when it explains a real invariant or ownership decision that future reviewers should know. |
| 40 | +- If `gh`/Gitcrawl reports `database disk image is malformed`, run `gitcrawl doctor --json` once to let the portable cache repair before retrying review; do not bypass the shim unless repair fails and freshness requires live GitHub. |
| 41 | +- If Gitcrawl reports a portable manifest mismatch, source/runtime DB health error, or stale portable-store checkout, run `gitcrawl doctor --json` and inspect `source_db_health`, `runtime_db_health`, and `portable_store_status` before falling back to live GitHub. |
| 42 | +- Do not push just to review. Push only when the user requested push/ship/PR update. |
| 43 | + |
| 44 | +## Pick Target |
| 45 | + |
| 46 | +Dirty local work: |
| 47 | + |
| 48 | +```bash |
| 49 | +<autoreview-helper> --mode local |
| 50 | +``` |
| 51 | + |
| 52 | +Use this only when the patch is actually unstaged/staged/untracked in the |
| 53 | +current checkout. For committed, pushed, or PR work, point the helper at the commit |
| 54 | +or branch diff instead; do not force `--mode local` / `--uncommitted` just |
| 55 | +because the helper docs mention dirty work first. A clean local review |
| 56 | +only proves there is no local patch. |
| 57 | + |
| 58 | +Branch/PR work: |
| 59 | + |
| 60 | +```bash |
| 61 | +<autoreview-helper> --mode branch --base origin/main |
| 62 | +``` |
| 63 | + |
| 64 | +Optional review context is first-class: |
| 65 | + |
| 66 | +```bash |
| 67 | +<autoreview-helper> --mode branch --base origin/main --prompt-file /tmp/review-notes.md --dataset /tmp/evidence.json |
| 68 | +``` |
| 69 | + |
| 70 | +If an open PR exists, use its actual base: |
| 71 | + |
| 72 | +```bash |
| 73 | +base=$(gh pr view --json baseRefName --jq .baseRefName) |
| 74 | +<autoreview-helper> --mode branch --base "origin/$base" |
| 75 | +``` |
| 76 | + |
| 77 | +Committed single change: |
| 78 | + |
| 79 | +```bash |
| 80 | +<autoreview-helper> --mode commit --commit HEAD |
| 81 | +``` |
| 82 | + |
| 83 | +or with the helper: |
| 84 | + |
| 85 | +```bash |
| 86 | +/Users/steipete/Projects/agent-scripts/skills/autoreview/scripts/autoreview --mode commit --commit HEAD |
| 87 | +``` |
| 88 | + |
| 89 | +Use commit review for already-landed or already-pushed work on `main`. Reviewing |
| 90 | +clean `main` against `origin/main` is usually an empty diff after push. For a |
| 91 | +small stack, review each commit explicitly or review the branch before merging |
| 92 | +with `--base`. |
| 93 | + |
| 94 | +## Parallel Closeout |
| 95 | + |
| 96 | +Format first if formatting can change line locations. Then it is OK to run tests and review in parallel: |
| 97 | + |
| 98 | +```bash |
| 99 | +scripts/autoreview --parallel-tests "<focused test command>" |
| 100 | +``` |
| 101 | + |
| 102 | +Tradeoff: tests may force code changes that stale the review. If tests or review lead to code edits, rerun the affected tests and rerun review until no accepted/actionable findings remain. Once that rerun exits cleanly, stop; do not spend another long review cycle on redundant confirmation. |
| 103 | + |
| 104 | +## Review Panels |
| 105 | + |
| 106 | +Run multiple reviewers against one frozen bundle: |
| 107 | + |
| 108 | +```bash |
| 109 | +<autoreview-helper> --reviewers codex,claude |
| 110 | +``` |
| 111 | + |
| 112 | +`--panel` is shorthand for Codex plus Claude unless `--engine` changes the first reviewer: |
| 113 | + |
| 114 | +```bash |
| 115 | +<autoreview-helper> --panel |
| 116 | +``` |
| 117 | + |
| 118 | +Set reviewer models and thinking/effort explicitly: |
| 119 | + |
| 120 | +```bash |
| 121 | +<autoreview-helper> --reviewers codex,claude --model codex=gpt-5.1 --thinking codex=high --model claude=sonnet --thinking claude=max |
| 122 | +``` |
| 123 | + |
| 124 | +Inline syntax is also supported: |
| 125 | + |
| 126 | +```bash |
| 127 | +<autoreview-helper> --reviewers codex:gpt-5.1:high,claude:sonnet:max |
| 128 | +``` |
| 129 | + |
| 130 | +Codex maps thinking to `model_reasoning_effort` and accepts `low`, `medium`, |
| 131 | +`high`, or `xhigh`. Claude maps thinking to `--effort` and also accepts `max`. |
| 132 | +Engines without a real thinking knob reject `--thinking`. |
| 133 | + |
| 134 | +## Context Efficiency |
| 135 | + |
| 136 | +Run the helper directly so target selection, engine choice, structured validation, and exit status all stay in one path. If output is noisy, summarize the completed helper output after it returns; do not ask another agent or reviewer to rerun the review. |
| 137 | + |
| 138 | +## Helper |
| 139 | + |
| 140 | +OpenClaw repo-local helper: |
| 141 | + |
| 142 | +```bash |
| 143 | +.agents/skills/autoreview/scripts/autoreview --help |
| 144 | +``` |
| 145 | + |
| 146 | +`agent-scripts` checkout helper: |
| 147 | + |
| 148 | +```bash |
| 149 | +skills/autoreview/scripts/autoreview --help |
| 150 | +``` |
| 151 | + |
| 152 | +Global helper from `agent-scripts`: |
| 153 | + |
| 154 | +```bash |
| 155 | +~/.codex/skills/agent-scripts/autoreview/scripts/autoreview --help |
| 156 | +``` |
| 157 | + |
| 158 | +If installed from `agent-scripts`, path is: |
| 159 | + |
| 160 | +```bash |
| 161 | +/Users/steipete/Projects/agent-scripts/skills/autoreview/scripts/autoreview --help |
| 162 | +``` |
| 163 | + |
| 164 | +The helper: |
| 165 | + |
| 166 | +- chooses dirty local changes first |
| 167 | +- otherwise uses current PR base if `gh pr view` works |
| 168 | +- otherwise uses `origin/main` for non-main branches |
| 169 | +- supports `--engine codex`, `claude`, `droid`, and `copilot`; default is `AUTOREVIEW_ENGINE` or `codex`; Codex should remain the default when nothing is set |
| 170 | +- use `--mode commit --commit <ref>` for already-committed work, especially clean `main` after landing |
| 171 | +- should be left in `--mode auto` or forced to `--mode branch` for PR/branch work; do not force `--mode local` after committing |
| 172 | +- writes only to stdout unless `--output`, `--json-output`, or live streamed engine stderr is set |
| 173 | +- supports `--dry-run`, `--parallel-tests`, `--prompt`, `--prompt-file`, `--dataset`, `--no-tools`, `--no-web-search`, and commit refs |
| 174 | +- supports `--stream-engine-output` or `AUTOREVIEW_STREAM_ENGINE_OUTPUT=1` for live engine text while preserving structured validation; Codex and Claude hide tool/file event details, emit compact activity summaries, and report usage at turn completion |
| 175 | +- supports opt-in review panels with `--panel` / `--reviewers`, plus per-engine `--model` and `--thinking` |
| 176 | +- allows read-only tools and web search by default where the selected CLI supports them; forbids nested review in the prompt; Codex is run through `codex exec` with read-only sandbox and structured output |
| 177 | +- prints `review still running: <engine> elapsed=<seconds>s pid=<pid>` to stderr at long-running intervals while waiting for the selected review engine, unless streamed output or compact Codex activity has been visible recently |
| 178 | +- prints `autoreview clean: no accepted/actionable findings reported` when the selected review command exits 0 |
| 179 | +- exits nonzero when accepted/actionable findings are present |
| 180 | + |
| 181 | +## Final Report |
| 182 | + |
| 183 | +Include: |
| 184 | + |
| 185 | +- review command used |
| 186 | +- tests/proof run |
| 187 | +- findings accepted/rejected, briefly why |
| 188 | +- the clean review result from the final helper/review run, or why a remaining finding was consciously rejected |
| 189 | + |
| 190 | +Do not run another review solely to improve the final report wording. If the final helper run exited 0 and produced no accepted/actionable findings, report that exact run as clean. |
0 commit comments