/aitask-review

Review code using configurable review guides, then create tasks from findings. This skill separates code quality review from implementation — first build something that works, then systematically review it for quality improvements.

Usage:

/aitask-review          # Interactive: choose scope and review guides

Note: Must be run from the project root directory. See Skills overview for details.

Codex CLI note: When continuing from this skill into implementation, in Codex wrappers, after implementation, most of the times you will need to explicitly tell the agent to continue the workflow because request_user_input is only available in plan mode. Example prompts: Good, now finish the workflow or Good, now continue.

Step-by-Step

1. Define review scope — Choose what to review: specific file paths, directories, or glob patterns; or select from recent commits (non-administrative commits are listed with pagination)
2. Select review guides — The skill auto-detects relevant guides from aireviewguides/ by analyzing target files (language, framework, project markers) and ranking guides by relevance. Guides are presented for multi-select with environment-specific matches first, then universal guides
3. Systematic review — For each selected guide, the skill reads the guide’s review instructions and examines the target code, recording findings with severity (high/medium/low), location, description, and suggested fix
4. Present findings — Results are grouped by review guide and severity. Select which findings to act on (individual selection or “select all”)
5. Create tasks — Three modes: a single task combining all findings, one task per review guide, or one task per finding. For multiple tasks, a parent + children structure is created automatically
6. Handoff — Choose to continue directly to implementation (via the standard /aitask-pick skill) or save the task(s) for later

Key Capabilities

• Auto-detection of relevant guides — The environment detection system scores review guides against your target files using project markers, file extensions, shebang lines, and directory patterns. You see the most relevant guides first, with an option to include non-matching guides
• Severity classification — Findings are categorized as high (security vulnerabilities, correctness bugs, data loss), medium (code quality, missing error handling, performance), or low (style, naming, cosmetics). Task priority is derived from the highest severity among selected findings
• Flexible task creation — Create a single task for quick fixes, group by guide for organized review rounds, or separate tasks for independent parallel work
• Execution profiles — Pre-configure which guides to auto-select via the review_default_modes profile key (comma-separated list of guide names). Set review_auto_continue to true to auto-proceed to implementation after task creation

Review Guides

Review guides are markdown files in aireviewguides/ organized by environment (e.g., general/, python/, shell/). Each guide has YAML frontmatter with metadata (name, description, reviewtype, reviewlabels, environment) and a body containing actionable review instructions.

To exclude specific guides from auto-detection, add patterns to aireviewguides/.reviewguidesignore. This file uses gitignore syntax:

general/performance.md    # Exclude a specific guide
android/                  # Exclude an entire environment
*.draft.md                # Exclude all draft guides
!general/security.md      # Re-include after a broader exclusion

To manage review guides, see the companion skills:

• /aitask-reviewguide-import — Import external coding standards and best practices as review guides
• /aitask-reviewguide-classify — Assign metadata to guides for better auto-detection and similarity matching
• /aitask-reviewguide-merge — Consolidate overlapping guides to avoid duplicate review checks

Workflows

For a full workflow guide covering the review cycle and guide management, see Code Review.