Skill: Handle an Octane GitHub issue

SkillDev tools

Lets your agent fetch a GitHub issue, triage it, and plan or implement a fix in the octane repo.

Use Skill: Handle an Octane GitHub issue in Claude, ChatGPT or Ahel Desktop

Free. Sign in, add Skill: Handle an Octane GitHub issue and connect your AI. About a minute.

Also: Claude Code · Cursor · Codex

Then ask your AI: use the Skill: Handle an Octane GitHub issue

Details

Instructions available. Your AI can read the instructions. Execution depends on the setup they require.

Add Ahel to your AI once: Claude, ChatGPT, Cursor, Claude Code or Codex. Then ask it to use this.

Skill: Handle an Octane GitHub issueStart free
About this skill

Work a GitHub issue in the octane repo. Use when given an issue number or link. Fetches the issue, classifies the owning area, and plans the fix.

What this skill tells your AI

The instructions your AI receives, as published by octanejs/octane in .agents/skills/handle-issue/SKILL.md and read by Ahel’s review.

Use this to inspect an issue, triage it, propose a solution, and optionally implement it.

Inputs

  • Issue number or URL.
  • Optional scope constraints: triage-only, propose-only, implement, or create PR.

Workflow

  1. Fetch issue context

    gh issue view <number> --json number,title,body,author,labels,state,comments,assignees,milestone,url
    

    If linked PRs/commits are mentioned, inspect them too.

  2. Classify

    • bug, feature, docs, test gap, performance, parity gap, ecosystem binding, question.
    • affected area: core runtime, compiler, SSR/hydration, Vite plugin, binding package, benchmarks/docs.
    • severity and likely user impact.
  3. Check project rules

    • Read AGENTS.md, docs/packages.md, and relevant docs.
    • For React-behavior issues, check docs/differences-from-react.md and classify intentional divergence vs bug.
  4. Reproduce or validate claim

    • Prefer a minimal failing test or fixture.
    • If not reproducible, document missing info and ask targeted questions.
    • Avoid broad rewrites before there is a failing test.
  5. Propose solution Include:

    • root-cause hypothesis
    • files likely to change
    • test plan
    • compatibility/divergence considerations
    • risk level
  6. Optional implementation

    • Follow bug-hunter.md for bugs.
    • Follow octane-react-library-port for binding/compat issues.
    • Follow octane-core-extend.md for runtime/compiler extensions.
  7. Issue response template

Thanks: I triaged this as <classification> affecting <area>.

Findings:
- ...

Likely cause:
- ...

Proposed fix:
- ...

Validation plan:
- ...

Notes:
- ...
  1. Labels/comments
    • Use gh issue edit/gh issue comment only when asked or when operating autonomously with permission.
    • Do not close issues without maintainer instruction unless explicitly authorized.
    • Classify issues with the existing bug/enhancement/documentation/question set. The feat/fix/docs/test/perf/refactor/chore/ci type labels are for PRs.
    • Add agent-authored to an issue an agent filed. When an agent only comments on someone else's issue, say so in the comment rather than relabelling their issue.
    • A PR that closes the issue is labelled by create-a-pr, not here.

Signals

GitHub stars
1k
Forks
61
Last commit
Oct 2026
Advanced
Item type
skill
Key
handle-issue
Source
github.com/octanejs/octane