richkuo/rk-skills/github-issue-format

Claude Code and agent skill by richkuo/rk-skills

Trust tier 2
  • community
LLM review pending
Source check pending

Checking that the SKILL.md file is still in the repository.

Required format for creating or editing any GitHub issue — [C<score>] title convention, complexity rationale line, complete-body rule, mandatory Plain simple English section, attribution footer, milestone title rule. Load BEFORE creating or editing a GitHub issue or milestone.

Install

npx vskill@latest install richkuo/rk-skills/github-issue-format

SKILL.md

GitHub issue format

Title

  • [C<score>] <title>: a plain-simple-English sentence in ASD-STE100, precise about component and behavior, e.g. [C95] Orders can be filled twice when two fills arrive at the same moment.
  • The complexity score (0–100) is a model and effort routing signal. validate-issue step 6 owns the formula, axes, and routing table; never restate or approximate them here.

Milestone

  • Every milestone title you create or rename is one hyphenated token with no spaces (MCP-Read-foundations, Backtest-live-parity). This holds for every path that creates one: a batch from prd-to-issues, a split of one issue into follow-ups, or a manual request.
  • Before you assign an issue to a milestone that has spaces in its title, rename that milestone to the hyphenated form (gh api -X PATCH repos/<owner>/<repo>/milestones/<number> -f title=<new-title>). The rename keeps every issue on the milestone. Quote a title that still has spaces in every gh call.

Body

  • Never create a placeholder, stub, or empty-bodied issue. Every issue gets a complete body at creation, even in a batch. If a follow-up is not ready to spec, track it in the parent issue or notes until it is.
  • Section order: complexity rationale line, ## Problem, ## Goal, ## Approach, ## Acceptance criteria, ## Plain simple English, then any Execution block, then the attribution footer.
  • Complexity rationale line (first line of the body): matches the title prefix and ends with an explicit plan signal:

**Complexity: 95/100** — Capability 3 (Risk 4, Uncertainty 2 — money/data-integrity on order-fill path); Volume 20 (Scope 4, Coupling 3, Verification 3) — Opus 5.5, high · plan: yes. The line always carries all five axis grades; the validate-issue scoring reference owns the axes and their anchors.

  • plan signal: · plan: yes when the score is ≥ 75 (a plan is posted before the build, by Opus 5.5 at xhigh when the score is ≥ 85 and at high below 85, unless a Plan model: Fable 5.1 stamp asks for Fable, whose plan runs only at high; the builder is Opus 5.5 at high); · plan: no below 75. Always write it. Compatibility (owned here): the signal and the Execution block key were once spelled · fableplan: yes|no and **fableplan first:**, and the PR-title suffix , fableplan. Every reader accepts both spellings, every writer emits only plan, and a restamp rewrites an old spelling to the new one. Fable 5.1 effort tiers follow the LLM Attribution Footer section of CLAUDE.md.
  • `## Plain simple English` is mandatory on every issue: one short paragraph under 55 words in ASD-STE100, per the CLAUDE.md/AGENTS.md Response Style definition. State what is wrong or missing and why it matters, for a reader who knows the product but not the code. Never restate the approach, list file paths or symbols, or give a time or effort estimate there. An edit that rewrites body prose adds the section when it is missing; an edit that changes only machine metadata (the Execution block, the [C<score>] title prefix, or the complexity rationale line, as a rescore-only edit does) leaves every prose section unchanged and does not add it.
  • LLM Attribution Footer ends the body: Created for a new issue, Validated when a validation pass produced the edit (validate-issue and its wrappers), Updated for any other edit. Stack the new line under the existing ones; never replace them, and never append an exact duplicate of a line already there.
  • Project precedence: a repo CLAUDE.md issue or footer format overrides this default.

Security

  • Security scanPassed, 100 of 100Pattern checks on the skill's files.
  • Model reviewNot runA model review is scheduled for the next rescan.

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
35
Repository forks
7
7-day trend score
18

Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/richkuo/rk-skills/github-issue-format/badge)](https://verified-skill.com/skills/richkuo/rk-skills/github-issue-format)

Something wrong with this skill or its page? Tell us.