Skip to content

generic-automation-and-it/smooth-ai-stockanalysis / .agents/skills/ai-review

ai-review

Verified

Analyze and execute AI PR review feedback with fix/skip decisions. Use when a user asks to parse an AI review, apply selected fixes, and finalize review processing for GitHub or Azure DevOps pull requests. Detects the review source — for a GitHub Copilot agent review it replies to and resolves each linked review thread; otherwise it appends AI review notes to the PR description.

FileKindSize
AGENTS.mdmarkdown4.1 KB
SKILL.mdskill md9.2 KB
agents/openai.yamlconfig201 B
scripts/copilot-review.shexecutablescript5.2 KB
# ai-review — Maintenance Context

## TL;DR

Vendored copy of the `/ai-review` consumer skill from `generic-automation-and-it/smooth-ai-report-review`. It *parses* an AI PR review and applies per-issue fix/skip decisions; it does **not** generate the review. Treat this directory as a mirror of upstream — fixes belong upstream, not here.

## Non-Negotiables

- **Do not hand-edit the skill logic here to fix bugs.** This is a downstream copy. Changes made locally diverge from upstream and are lost on the next sync. File the fix in the upstream repo, then re-copy.
- **Only the `ai-review` consumer skill is installed.** The review *generator* (`ai-review-report` skill + its scripts) is intentionally NOT copied — it runs remotely via the reusable GitHub Actions workflow (`.github/workflows/pipeline-code-review-report.yml`, which `uses:` the upstream `pipeline-code-review-report.yml@main`). Do not copy the generator tree in to "complete" the install.
- **Script-path duality:** `SKILL.md` references `.agents/skills/ai-review/scripts/copilot-review.sh` for this copy-install. The `${CLAUDE_PLUGIN_ROOT}/...` variant in `SKILL.md` applies only when the skill runs from the upstream Claude Code plugin (`smooth-ai-review`) — not here. Keep the copy-install path permitted in `.agents/settings.json`.

## System Context

```mermaid
C4Context
  System(consumer, "ai-review (this skill)", "Parses a posted review; applies fix/skip; routes results")
  System_Ext(gha, "pipeline-code-review-report.yml@main", "Remote reusable GHA that GENERATES the review on a PR")
  System_Ext(analyse, "pipeline-ai-analyse.yml", "Companion low/medium self-fix loop")
  System_Ext(gh, "GitHub PR", "Review threads / PR description / comments")
  Rel(gha, gh, "posts AI review")
  Rel(analyse, gh, "reads latest gate review, pushes safe fixes, posts summary")
  Rel(consumer, gh, "reads review, replies/resolves threads or appends AI Review Notes")
```

Two halves, deliberately split: the remote GHA produces the review report on a PR; this local skill consumes it. The user drives the consumer with `/ai-review <pr>` (analyse) then `/ai-review <pr> 1=fix 2=skip` (execute).

The companion `.github/workflows/pipeline-ai-analyse.yml` runs after `PR Code Review Report` completes. Its guard reads the latest gate-authored review, extracts only 🟡 Medium and 🔵 Low findings, invokes fetched upstream `ai-analyse` tooling, pushes any safe auto-fixes, and posts an `ai-analyse auto-fix summary`. The loop is bounded by `OPENCODE_ANALYSE_MAX_INCREMENTAL` so repeated incremental reviews cannot run forever.

## Key Behaviors

- **Mode auto-detect:** any `N=fix`/`N=skip` arg → execute; otherwise analyse. Analyse always STOPS and never proceeds to execute on its own.
- **Source routing:** auto-detects Copilot vs other via `copilot-review.sh detect <pr>`. Copilot → reply/resolve each linked review thread + post a summary comment. Other → append the fix/skip table to the PR description's AI Review Notes (append, never overwrite).
- **All deterministic GitHub plumbing** (detect/threads/reply/resolve/summary) lives in `scripts/copilot-review.sh`; the skill keeps only the judgment (parsing + fix/skip + reply text).

## Changelog

| Date | Change | Ref |
|:-----|:-------|:----|
| 2026-06-20 | Vendored `/ai-review` consumer skill from smooth-ai-report-review; generator kept remote via thin caller workflow. | |
| 2026-07-05 | Documented the companion self-fix workflow that consumes low/medium review findings and posts an auto-fix summary. | |
| 2026-07-23 | Vendored the consumer skill and companion workflows into `smooth-ai-stockanalysis`. | #248 |
| 2026-07-23 | Local delta from upstream: reworded the analyse-mode guardrail for explicit human-in-the-loop clarity. Re-apply on next upstream sync. | #248 |
| 2026-07-24 | Local delta from upstream: tightened Execute step 5 and the matching Guardrail to make the `ci: /ai-review — processed review responses` empty commit MANDATORY whenever the review contains at least one 🔴 Critical or 🟠 High finding (fix OR skip). Aligns the vendored skill with the root AGENTS.md non-negotiable. Re-apply on next upstream sync. | #252 |
writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deploymentStars: 323kUpdated Claude CodeCodexVerifiedInstall
xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release.Stars: 118kUpdated Claude CodeCodexVerifiedInstall