codecanary-loop

分类: 开发与编程 | 上传者: alansikora | 下载: 0 | 版本: v1.0（最新）

推动代码金丝雀审查→分类→修复→推动反馈循环收敛。默认为 PR 模式（观察 codecanary GitHub 操作、获取结果、应用批准的修复、提交、推送和重新观察）。当未检测到 PR 时，自动回退到本地模式，检查未提交的更改并跳过所有 git 管道。在应用之前始终与用户确认每项发现——从不自动应用。 --- # codecanary-loop 您正在驱动 CodeCanary 审查修复循环。当操作员希望您迭代 CodeCanary 的发现直到审核干净为止时，他们会运行此技能。保持纪律：您是 CLI 和操作员决策之间的粘合剂，而不是审核者。 ## CLI 中的繁重工作所有轮询、获取、解析和 PR/repo 自动检测都发生在“codecanary discovery”和“codecanary review”中。你永远不会直接从这个技能中掏出“gh”。您永远不会解析 HTML 注释标记。您永远不会轮询 CI 状态。 CLI 发出结构化 JSON；你消耗它。

更新日志: Source: GitHub https://github.com/alansikora/codecanary

目录结构

当前层级: tree/main/.claude/skills/codecanary-loop/

📄 SKILL.md 7.0 KB

SKILL.md

---
name: codecanary-loop
description: |
  Drive a codecanary review → triage → fix → push feedback loop to convergence.
  Defaults to PR mode (watches the codecanary GitHub action, fetches findings,
  applies approved fixes, commits, pushes, and re-watches). Falls back to
  local mode automatically when no PR is detected, reviewing uncommitted
  changes and skipping all git plumbing. Always confirms every finding with
  the user before applying — never auto-applies.
---

# codecanary-loop

You are driving the CodeCanary review-fix loop. The operator runs this skill
when they want you to iterate against CodeCanary's findings until the review
is clean. Stay disciplined: you are the glue between the CLI and the
operator's decisions, not the reviewer.

## Heavy lifting lives in the CLI

All polling, fetching, parsing, and PR/repo autodetection happens in
`codecanary findings` and `codecanary review`. You never shell out to `gh`
directly from this skill. You never parse HTML comment markers. You never
poll for CI status. The CLI emits structured JSON; you consume it.

This is intentional for token-efficiency: the loop machinery runs in
subprocesses whose output is small and structured. Your conversation budget
is spent on triage judgment and fix application, not on watching CI.

## Mode selection

- **PR mode (default)**: fixes land as commits on the current branch and
  are pushed. Used when an open PR exists for the branch.
- **Local mode** — activated automatically when no PR is detected for the
  current branch: `codecanary review --output json` runs a review on the
  current dirty working tree; fixes are applied but not committed or pushed.

If you cannot tell which mode applies, ask the operator before starting.

## The loop

Track one piece of state across iterations:
- `CYCLE` — integer, starts at 0, increments at the top of every iteration.

### Iteration

1. `CYCLE = CYCLE + 1`.
2. Fetch findings:
   - **PR mode**: run
     `codecanary findings --watch --output json`.
     The command blocks until the review check completes; its stdout is
     a single JSON object. Parse it. Deduplication is handled by GitHub
     thread resolution — resolved threads are excluded by default.
     Note: findings the operator previously skipped (via "Skip this
     cycle" or "Apply some") will re-appear if their threads are still
     open. This is intentional — skipped findings are deferred, not
     dismissed.
   - **Local mode**: run `codecanary review --output json`. The command
     runs the review inline; its stdout is a JSON object with a
     `findings` array in the same shape.
3. **PR mode only** — check the `conclusion` field in the JSON output.
   (Skip this step entirely for local mode — there is no check run.)
   If `conclusion` is `failure`, the review run itself broke. If
   `conclusion` is `cancelled` or `timed_out`, the run was interrupted
   (e.g. a newer push superseded it). In any of these cases — or any
   value other than `success` / `neutral` / empty — tell the operator
   the check failed, name the conclusion, and stop. Do not say the
   review is clean, even if `findings` is empty — an empty list on a
   failed run means findings were never published, not that the code
   is fine. Roll `CYCLE` back by one (`CYCLE = CYCLE - 1`) so the
   next retry starts at the correct count. Wait for the operator to
   explicitly ask you to retry before starting another cycle.
4. If the findings list is empty (for either mode), tell the operator
   the review is clean and exit. Do not loop further.
5. If `CYCLE > 1`, emit this reminder to the operator before the
   triage table, substituting *N* with the current value of `CYCLE`:
   > This is review cycle *N*. Before applying fixes, check whether the new
   > findings are caused by your previous fixes or are genuinely different
   > issues. If the bot keeps re-flagging the same `fix_ref` across cycles,
   > stop and verify your fix actually addresses what the bot meant —
   > don't keep patching symptoms.
6. Render a triage table (Markdown) summarizing the findings:
   - Columns: severity, file:line, fix_ref, title, proposed action
   - One row per finding. Keep proposed actions terse (one line each).
7. Ask the operator to confirm. Use `AskUserQuestion` with a single
   question whose options are:
   - "Apply all" *(Recommended)*
   - "Apply some (I'll specify which)"
   - "Skip this cycle" — treats all findings as deferred; exits the loop
   - "Abort" — exits the loop immediately
   Wait for the response before touching any files.
8. If the operator approved (all or some), apply the fixes. For each
   approved finding:
   - Read the file, make the minimal edit that addresses the finding,
     keeping the surrounding code intact (do not "improve" unrelated code).
   - If the suggestion in the finding is an exact code snippet and fits
     the context, prefer it verbatim; otherwise adapt it to the codebase
     conventions (existing imports, types, error-handling style).
9. Finalize the cycle:
   - **PR mode**:
     - Run `go build ./...` and `go test ./...` if any Go files changed.
     - Commit with a message like:
       `fix: address codecanary review on #<PR> (cycle <N>)`
       plus a brief bullet list of which findings were addressed.
     - Push the branch.
     - Go back to step 1.
   - **Local mode**: stop. Report the summary of applied fixes to the
     operator. Do not commit, do not push, do not loop — a single pass
     is the contract for local mode.

## Stopping conditions

Exit the loop (and tell the operator *why*) whenever any of these hold:

- **PR mode**: the findings list comes back empty and `conclusion` is
  healthy (`success` or `neutral`) — normal success.
- **Local mode**: the findings list comes back empty — normal success.
  (There is no `conclusion` field in local mode; its absence is expected.)
- The operator chose "Skip this cycle" or "Abort".
- The CLI errors out (network failure, no PR detected, timeout on
  `--watch`). Surface the error verbatim and stop.
- You detect you're in a stable disagreement loop: the same `fix_ref`
  values appear in two consecutive cycles after you applied fixes for
  them. This is the signal from step 5 turning into a hard stop — tell
  the operator which fix_refs keep re-emerging and ask them to review
  whether the fix is correct before continuing.

## What not to do

- Don't iterate without operator confirmation.
- Don't auto-apply nitpicks or "obvious" fixes.
- Don't write your own logic to parse `<!-- codecanary:finding ... -->`
  markers — the CLI already returns structured Findings.
- Don't `gh api` or `gh pr view` yourself — the CLI handles that.
- Don't attempt concurrent PR work. One branch at a time.
- Don't commit to `main` or an unrelated branch; always stay on the PR's
  feature branch.
- Don't force-push. The loop only appends commits.

## Example operator turn

```
user: Run codecanary-loop on this PR

assistant: (invokes `codecanary findings --watch --output json`,
            parses JSON, renders triage table, asks for confirmation,
            applies approved fixes, commits, pushes, loops)
```

登录后下载/点赞/收藏 ❤ 10 | ★ 0

codecanary-loop

目录结构

SKILL.md

举报内容

提示