Skill Gallery
← Back to list
silver2dream

run-issues

by silver2dream

AI-assisted development workflow kit for Spec → Implement → PR → Merge automation

⭐ 0🍴 0📅 Jan 24, 2026
ai-workflow-automationclaude-codecodex-clideveloper-toolsgithub-actionsspec-driven
View on GitHubRun in Manus

SKILL.md

name: run-issues description: Autonomous issue management - fetches open GitHub issues, analyzes priorities, detects dependencies, spawns parallel subagents, and verifies results. Triggers: /run-issues, autonomous issues, batch issues, parallel issue processing. allowed-tools: Read, Grep, Glob, Bash, Task

Run Issues Skill

Autonomous workflow for processing multiple GitHub issues in parallel with priority-based scheduling and verification.

Overview

This skill orchestrates the complete lifecycle of issue processing:

  1. Fetch open issues from GitHub
  2. Analyze and prioritize (P0 > P1 > P2)
  3. Detect dependencies between issues
  4. Spawn subagents for parallel execution
  5. Verify all results before reporting

When to Use

Use this skill when:

  • User invokes /run-issues
  • User requests batch processing of issues
  • User wants autonomous issue handling
  • Multiple issues need parallel processing

Workflow

Phase 0: Pre-Flight Check

IMPORTANT: Before starting, check if AWK principal workflow is active.

# Check for active AWK workflow
if [ -f ".ai/state/kickoff.lock" ]; then
    echo "⚠️ WARNING: AWK principal workflow is active"
fi

If .ai/state/kickoff.lock exists:

  1. Warn the user: 
    ⚠️ AWK principal workflow 正在執行中。
同時執行 /run-issues 可能導致：
- 同一 issue 被重複處理
- 產生重複的 branch 或 PR
- Merge conflicts

建議：等待 AWK workflow 完成後再執行。
確定要繼續嗎？(yes/no)
  2. Only proceed if user explicitly confirms with "yes"
  3. If user says "no", abort gracefully

If lock file does not exist, proceed to Phase 1.

Phase 1: Fetch Issues

gh issue list --state open --json number,title,body,labels,assignees --limit 50

Parse the JSON output to get all open issues.

Phase 2: Analyze Issues

Read phases/analyze.md for detailed priority analysis rules.

For each issue:

  1. Extract priority from labels (P0/P1/P2)
  2. Parse dependencies from issue body
  3. Calculate priority score
  4. Build dependency graph

Phase 3: Plan Parallel Execution

Read phases/parallelize.md for parallelization strategy.

  1. Group issues by dependency chains
  2. Identify independent issue sets
  3. Determine optimal subagent count (max 3 concurrent)
  4. Create execution batches

Phase 4: Execute

For each batch of independent issues:

  1. Spawn subagents using Task tool:

    Task tool parameters:
- subagent_type: general-purpose
- description: "Work on Issue #<number>: <title>"
- prompt: |
    Complete Issue #<number>. Requirements: <body excerpt>.

    CRITICAL - Follow .ai/rules/_kit/git-workflow.md strictly:

    Commit format:
    - Format: [type] subject
    - Subject MUST be lowercase (e.g., "add feature" not "Add feature")
    - NO colon after bracket
    - Valid types: feat, fix, docs, style, refactor, perf, test, chore

    Examples:
    ✅ [docs] update api reference
    ✅ [feat] add user authentication
    ❌ [Docs] Update API reference (uppercase = WRONG)
    ❌ docs: update api reference (colon = WRONG)

    PR requirements:
    - PR body MUST include: Closes #<number>
    - PR target: develop branch (or as specified)

  2. Wait for all subagents in batch to complete

  3. Move to next batch (respecting dependencies)

Phase 5: Verify Results

Read phases/verify.md for verification checklist.

For each completed issue:

  1. Check if PR was created
  2. Validate commit format
  3. Verify tests pass
  4. Confirm branch is up-to-date

Phase 6: Report

Generate summary table:

| Issue | Title | Priority | Status | PR | Notes |
|-------|-------|----------|--------|-----|-------|
| #123  | Fix bug | P0 | Completed | #456 | Tests pass |
| #124  | Add feature | P1 | Completed | #457 | Ready for review |
| #125  | Update docs | P2 | Failed | - | Test failures |

Error Handling

  • If a subagent fails, log the error and continue with other issues
  • Failed issues are marked in the final report
  • Dependencies of failed issues are skipped with note

Self-Check

On each phase entry, output:

[RUN-ISSUES] <timestamp> | <phase> | loaded: <filename>

Integration with AWK Workflow

⚠️ WARNING: Do NOT run /run-issues and awkit kickoff simultaneously.

This skill processes GitHub issues independently of AWK's spec-driven workflow.

  • AWK uses tasks.md specs → awkit dispatch-worker
  • This skill fetches from GitHub → Task subagents

Recommended workflow:

/create-issues → issues created
        ↓
    Choose ONE:
        ├─ awkit kickoff (structured, spec-driven)
        └─ /run-issues (autonomous, batch processing)  ← this skill

Quick Reference

PhaseActionFile
Pre-FlightCheck AWK workflow not active(inline)
AnalyzeDetermine priorities and dependenciesphases/analyze.md
ParallelizePlan execution batchesphases/parallelize.md
VerifyCheck completed workphases/verify.md

Score

Total Score

65/100

Based on repository quality metrics

✓SKILL.md

SKILL.mdファイルが含まれている

+20
✓LICENSE

ライセンスが設定されている

+10
○説明文

100文字以上の説明がある

0/10
○人気

GitHub Stars 100以上

0/15
✓最近の活動

1ヶ月以内に更新

+10
○フォーク

10回以上フォークされている

0/5
✓Issue管理

オープンIssueが50未満

+5
✓言語

プログラミング言語が設定されている

+5
✓タグ

1つ以上のタグが設定されている

+5

Reviews

💬

Reviews coming soon