skills/issue-creator/SKILL.md
Turn a one-line request into a schema-compliant GitHub issue draft or creation command. Use when the user gives a short bug report, feature request, refactor idea, or task summary and wants a normalized issue with title, type, severity, description, acceptance criteria, and verification steps.
npx skillsauth add Cheggin/skill-chain issue-creatorInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Turn a short request into a valid GitHub issue for this repo.
Before drafting or creating an issue, read these files:
.harness/issue-schema.md.github/ISSUE_TEMPLATE/feature.yml.github/ISSUE_TEMPLATE/bug.ymlIf you will actually run gh issue create or validate a command, also read:
.claude/hooks/validate-issue-create.mjsDo not rely on memory. When sources disagree, use this precedence:
.claude/hooks/validate-issue-create.mjs for what the CLI hook will accept.harness/issue-schema.md for the normalized body shape.github/ISSUE_TEMPLATE/*.yml for wording and field intentImportant:
[feat], [fix], [docs].gh issue create when the user explicitly asks you to create the issue on GitHub.Given a one-liner, produce either:
gh issue create command that passes the local issue validatorThe output must be specific enough that another agent can pick it up without additional clarification.
Affected Packages, title wording, estimate, and agent category.TBD after repo scan instead of guessing.Choose exactly one:
fix: broken behavior, incorrect output, failure, regression, flaky behaviorfeat: new capability, new workflow, new surface arearefactor: internal cleanup without changing external behaviortest: missing or improved automated coveragedocs: documentation-only workchore: maintenance, cleanup, dependency or housekeeping workperf: performance improvementci: pipeline, automation, or workflow config changesWhen in doubt:
fixfeatdocsci or choreChoose one:
P0: system broken, blocks all work, deploy/build/data-loss level problemP1: feature broken or output unusable, but there is a workaroundP2: enhancement, missing guardrail, quality improvement, normal bug with limited blast radiusP3: low-impact polish, docs, or nice-to-have cleanupSafe default: P2.
Never use P0 or P1 unless the one-liner clearly describes breakage.
TBD after repo scan.Use a rough size:
XS: tiny copy or config tweakS: localized change in one areaM: several related files or one new skill/workflowL: cross-package or multi-surface changeXL: large multi-phase effort that likely should be splitIf the request looks XL, prefer splitting into multiple issues.
Choose the primary owner:
coding: product code, skills, packages, hooks, agentscontent: docs, copy, content artifactsgrowth: marketing, SEO, analytics, acquisitionoperations: deploy, infra, CI, monitoring, dependency upkeeporchestration: harness flow, coordination, state managementquality: tests, evals, audits, validationUse this exact structure:
## Type
<type>
## Severity
<P0|P1|P2|P3>
## Description
<what needs to change and why; focus on outcome, not implementation trivia>
## Affected Packages
<comma-separated paths or TBD after repo scan>
## Acceptance Criteria
- [ ] <machine-verifiable outcome 1>
- [ ] <machine-verifiable outcome 2>
- [ ] <machine-verifiable outcome 3>
## Verification Steps
1. <step that proves criterion 1>
2. <step that proves criterion 2>
3. <step that proves criterion 3>
## Estimate
<XS|S|M|L|XL>
## Agent Category
<coding|content|growth|operations|orchestration|quality>
Use this title format exactly:
[type] Imperative summary
Example:
[feat] Add issue-creator skill for normalized GitHub tickets
When the one-liner describes a bug:
## Type as fix## Description to capture both the broken behavior and the expected outcomeEvery checkbox must be testable by another agent. Good criteria mention observable outcomes such as:
Avoid vague criteria such as:
If the user asked for a draft, return:
Do not call GitHub unless the user explicitly asked you to post the issue.
If the user wants the issue posted, use gh issue create with a heredoc body so the local hook can parse it reliably.
The current validator accepts either <<EOF or <<'EOF'. Prefer <<EOF for consistency with the examples below.
gh issue create \
--title "[type] Imperative summary" \
--body "$(cat <<EOF
## Type
type
## Severity
P2
## Description
...
## Affected Packages
...
## Acceptance Criteria
- [ ] ...
## Verification Steps
1. ...
## Estimate
M
## Agent Category
coding
EOF
)"
Before you finish, verify:
[type]P0 through P3## Description## Acceptance Criteria## Verification StepsP0 or P1gh issue create before checking the local validator shapedevelopment
Design engineering principles for making interfaces feel polished. Use when building UI components, reviewing frontend code, implementing animations, hover states, shadows, borders, typography, micro-interactions, enter/exit animations, or any visual detail work. Triggers on UI polish, design details, "make it feel better", "feels off", stagger animations, border radius, optical alignment, font smoothing, tabular numbers, image outlines, box shadows.
documentation
Agentic memory system for writers - track characters, relationships, scenes, and themes
documentation
LLM Wiki — persistent markdown knowledge base that compounds across sessions (Karpathy model)
development
Build production-quality SaaS websites with opinionated design presets. Use when creating any startup website. The user MUST pick a design style before building. Enforces shadcn/ui, Figma design principles, specific CSS values per style, and anti-AI-writing. Load alongside anti-ai-writing skill. LIGHT MODE ONLY.