loopx-pr-review

# LoopX PR Review This skill is a thin host adapter. The built-in `pull-request-review` capability owns review depth, evidence requirements, completeness, and verdict policy through the CLI packet. Do not copy those rules into this skill or replace them with a host-specific checklist. ## Route Use this skill for `/loopx-pr-review`, explicit PR reviews, or review queues by state or time window. Route approval, merge, self-merge, and admin bypass to `loopx-pr-merge` (optional repo-kept workflow, not installed by default) after the evidence review is complete; it never replaces this skill's exact-head gate. Run `loopx --format json pr-review --state all` before ad hoc GitHub reads. Translate only explicit filters: - `--repo owner/repo` - `--since ISO` - `--state open|merged|all` - `--limit N` - `--review-priority other-developers-first|owner-first` (default `other-developers-first`; use `owner-first` to opt into owner priority) When omitted, the CLI resolves `pull_request_review` from the standard machine capability editor; an absent namespace keeps the default `other-developers-first`. Words such as `today`, `open`, or `merged` are filters, not permission to return a table only; stats-only output needs an explicit opt-out such as `只统计` or `stats only`. ## Preserve The Packet Save the full first JSON packet before printing a compact projection. Keep all paths named by `agent_response_contract.required_packet_fields_to_preserve`: - `agent_response_contract.review_execution_contract` - `result_completeness` and `scheduling_policy` - `review_groups` - `pull_requests[review_action_kind!=null].review_plan` - `pull_requests[review_action_kind!=null].review_template` - `pull_requests[review_action_kind!=null].evidence_commands` Do not pipe the only copy through `jq`. When an exhaustive request has `result_completeness.complete=false`, rerun with its `recommended_limit` before reviewing. Require the revision the installed capability declares, not a literal this file pins: read `review_execution_contract.policy_revision` from the packet and require the result's `review_policy_revision` to equal it. A schema name alone is insufficient. If missing or unequal, do not publish APPROVE; a conservative REQUEST_CHANGES is allowed only when it names the incompatible-policy gap. Do not retain expired temporary worktree overrides. Honor explicit runtime pins, but report incompatible policy rather than silently downgrading the review. ## Read The Review Frame First Read the PR's existing comments and cited documents first: a maintainer comment names the contract the change is judged against. - Review inside that frame; cite the document and check its rows, gates, or exit criteria one by one at the exact head, attaching each finding to its row. - Name unsatisfied rows as missing frame items with their own evidence, and keep the declared boundary (preview versus promotion, milestone entry versus merge). - Publish the frame-aligned conclusion as a PR comment citing that guidance, since a published review body cannot be edited. ## Execute One Review Plan Follow `scheduling_policy` and its ranked actionable `review_sequence`; explicit current-request PR selection may override ordering only, never `pull_requests[].review_action_kind` or exact-head idempotency. Generic `re-review`, `重新review`, and `复审` wording selects the named PR; it is not a force-refresh token. Todo/monitor prose may not select work. When `review_action_kind` is null, the row stays in `pull_requests` inventory but must not appear in `review_sequence`; its `review_plan` and `review_template` are null and `evidence_commands` is empty. Do one compact exact-head conclusion readback and report the existing verdict or bounded invalid/missing reason. Run a fresh audit only when the user explicitly requests fresh evidence despite that no-action result, or supplies a concrete new concern/evidence invalidation; regenerate with `--fresh-audit-exact-head NUMBER@HEAD_OID`, then execute the complete current plan and never inherit the earlier approval. For every actionable PR: 1. Record the packet's exact head. Follow `review_execution_contract.decision_procedure`, starting with the current goal and its delivery judgment in `problem_context`, including on re-review; then run `evidence_commands` and relevant repository-native validation. 2. Fill `review_plan.result_template` from the shared execution contract; preserve missing evidence as `unverified`. Execute its repository-reuse, default-off, authority and real-path counterfactual requirements rather than repeating them as prose. Never infer `verified` from metadata or CI. 3. Apply `completion_gate` literally. Save the filled result and check it before publication: ```bash loopx --format json pr-review --check-result review-result.json --packet review-packet.json ``` Fix contradictory verdicts, not evidence labels to obtain a pass. This local check cannot verify evidence truth, architecture judgment, remote freshness, or an old result relabeled without executing the current plan. Verified rows fill their declared fields; validation rows bind typed `case_id` coverage, and missing material evidence needs a concrete request-changes reason. 4. Render the verified result through `review_template`. The five sections are output structure, while the execution contract is the evidence authority. 5. Re-read the remote head immediately before verdict and publication. Restart the evidence pass if it changed. Each PR gets an independent evidence pass and standalone card; a queue table is only a preface. Finish fewer complete cards rather than metadata-only reviews. For managed review, pass `--goal-id GOAL` and follow the packet’s resolved `wait_for_ci`: false means never fetch, poll, or wait for CI; true retains CI validation. Required local failures/skips always block. Configure one Goal with `configure-goal --goal-id GOAL --no-pr-review-wait-for-ci --execute`; clear with `--clear-pr-review-configuration --execute`. ## Publish And Read Back For an open PR, publish validated actionable findings by default unless the user requested local-only/dry-run output or the finding is private or security-sensitive. - Remaining blocker: formal `REQUEST_CHANGES`; for an author-owned PR, use a `COMMENTED` review titled `Request changes conclusion (author-owned PR; GitHub blocks formal self-review)`. - Non-blocking finding with no blockers: formal `APPROVE`, not a bare comment; when the account is the author and self-approval is rejected, record the same conclusion as a `COMMENTED` review titled `Approval conclusion (author-owned PR; GitHub blocks formal self-approval)`. - Non-blocking P2 suggestions: still `APPROVE`; keep them in the body. - Merged PR: publish a post-merge audit comment only for a new actionable finding; avoid duplicating an equivalent exact-head result. Build public text from the exact reviewed head. Remove local paths, private context, raw logs, credentials, and internal-only links. Read the published review back, verify its state and rendered body, and return its URL. Merge still routes through `loopx-pr-merge`; an `APPROVE` is not merge authority. Do not leave a public blocker only in chat. Immediately before every merge, run `loopx --format json pr-review --repo OWNER/REPO --check-merge-readiness NUMBER@HEAD_OID`; merge only when it returns `ready=true` for that unchanged head. A rebase/update restarts review, admin bypass never overrides this gate, and author-owned fallback needs user authority. ## Full PR Review And Bilingual Format Every review must cover the whole PR, not only the top finding: read the full diff/local validation, then explain motivation, architecture, changed symbols, both paths, whole-diff risk, validation, and judgment. A findings-only or blocker-only body is incomplete. Publish two artifacts: 1. **详细中文评审** - a standalone Chinese full-PR review with the exact head and five sections: `动机`, `改动思路`, `具体改动`, `对主干的风险`, `我的整体评价`. Cover every changed surface and key symbols, not only the main finding. 2. **英文简短结论** - a line starting with exactly `English verdict:` followed by the bare token `APPROVE` or `REQUEST_CHANGES`, then the head, key finding, and validation. The parser matches that token as a keyword, so `English verdict: Approved at ...` is invalid and leaves the head without a merge-ready approval; write `English verdict: APPROVE - ...`. Do not publish before the Chinese section covers the entire PR. Read both artifacts back. ## Example / Walkthrough / Smoke-Only PRs When the review plan marks `smoke_or_example_only`, the `durable_smoke_value` evidence is mandatory before approval. The essence is real, durable value to the repository and product: running, deterministic, and public-safe are necessary but not enough. 1. Name the shipped behavior, boundary, or maintenance cost this artifact guards. "Demonstrates something that already works" is not durable value. 2. Scan existing coverage (`rg -l '<behavior|module>' examples tests`) and the same-author batch (`gh pr list ... --author <author>` / `gh search prs`); flag same-shape batches opened within minutes as PR farming. 3. Apply the repo smoke policy: thin + durable, guard shipped behavior or a real boundary, compress rather than append, consolidate same-shape walkthroughs into one PR or focused tests. 4. Verdict: `REQUEST_CHANGES` for duplicative, oversized, or value-less scaffolding; name the consolidation or thinning repair in the body. 5. Repeat offenders: after a REQUEST_CHANGES warning, further low-value same-shape PRs from the same author escalate to a contribution-restriction recommendation (owner blocks the account from further PR submissions); the warning must name this consequence. ## Autonomous Queue For recurring observation, keep one ignored checkpoint and use `loopx --format json pr-review --repo owner/repo --state open --autonomous-observation --observation-state-file .local/pr-review-monitor.json` with the projected or handled exact-head flags when those durable receipts exist. Treat `candidate` as a preview, not a durable projection. Follow this order: durable Todo target-key readback -> `--projected-exact-head` -> exact-head review/comment readback -> `--handled-exact-head`. Never send the projection ACK before the Todo exists, or the handled ACK before readback at that head. Observation states remain literal; the checkpoint grants no authority. Stateless callers may use `--previous-observation-json` instead. ## Failure If `loopx pr-review` is unavailable, repair the LoopX install or use the intended checked-out CLI. Do not reconstruct the queue manually and call it a successful `/loopx-pr-review` run.