opencode-skill-creator
Create, test, evaluate, and iterate OpenCode skills with mandatory intake interview.
- License 许可
- Apache-2.0
- 最近更新
- 2026-11-10
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- opencode-skill-creator
- triggers
- create skill, edit skill, opencode skill, skill eval, evaluate prompt, benchmark skill, iterate skill, skill testing, skill creation
- changelog
- docs/ciclos/cycle28-20260815.md
- description
- Create, test, evaluate, and iterate OpenCode skills with mandatory intake interview.
- token_budget
- 2700
02 资源文件 7
- 其他.opencode-skill-creator-version7 B
- 参考文档agents/analyzer.md10.1 KB
- 参考文档agents/comparator.md7.1 KB
- 参考文档agents/grader.md8.9 KB
- 参考文档references/description-optimization.md547 B
- 参考文档references/schemas.md11.8 KB
- 参考文档templates/eval-review.html6.9 KB
03 SKILL.md 正文
When to Use
Create, test, evaluate, and iterate OpenCode skills with man
Loop: Decide → Draft → Test prompts → Run (with-skill & baseline) → Review → Iterate → Install.
Creating a skill
Intake (mandatory): 3-5 questions on behavior, triggers, output quality, workflow, test cases. Skip only if confirmed (warn once). Summarize before drafting.
Write SKILL.md in %TEMP%/opencode-skills/<name>/. Name: ^[a-z0-9]+(-[a-z0-9]+)*$. Description explicit + "pushy" (OpenCode undertriggers). All trigger info in frontmatter. Structure: <name>/{SKILL.md, scripts/, references/, assets/}. Keep <500 lines, ref files >300L need TOC.
Test Cases: 2-3 realistic prompts, user sign-off, save to evals/evals.json.
Iteration: Apply → re-run into iteration-N+1/ → skill_serve_review with previousWorkspace → repeat until done. Don't overfit. Prefer explanation over MUST/ALWAYS.
Installation: Copy to .opencode/skills/<name>/ (project) or ~/.config/opencode/skills/<name>/ (global). Validate with skill_validate.
Plugin tools: skill_validate · skill_parse · skill_eval · skill_aggregate_benchmark · skill_serve_review · skill_stop_review
Anti-Rationalization
| Rationalization | Red Flag | Verification |
|---|---|---|
| "One skill fits all" | Skill with >3 responsibilities | Skill = single capability (SRP) — split, don't bloat |
| "Copy-paste is faster" | Creating skill without intake interview | Mandatory intake: trigger × description × example × budget (spec 5-4-09) |
| "Testing can wait" | Skill without iteration benchmark | skill_serve_review + iteration-N+1/ until done before ship |
Red Flags
- Skill description contains
MUST/ALWAYSwithout explanation → overfit (prefer explanation over MUST) - Token budget not bumped after adding capability → BLOCKER at regression gate — reject ship until budget is re-measured
Verification
skill_validate+skill_aggregate_benchmarkpass before publish- Test skill via subagent with real task before marking complete
Refs: opencode-skill-creator · skill-testing · skill-registry · skill-improver · karpathy-loop
docs/skills/opencode-skill-creator/reference.md
04 同仓库更多 Skills
web-quality-audit
Comprehensive web audit: performance, a11y, SEO, responsive, animation, design tokens.
ui-engine
UI system — Grid/Flexbox/@layer/:has(), container queries, compositor-only animation, OKLCH tokens, component patterns
baseline-ui
Anti-slop UI — layout, typography, responsive, animation, tokens. Use for cleanup or polish.
skill-creator
Create new AI agent skills following Agent Skills spec -- auto-bootstrap unknown tasks, define structure, register in AGENTS.md
senior-engineer
Apply senior engineering competencies — system design, trade-off analysis, delegation, mentoring, and tech debt strategy
prompt-engineering
Design effective prompts using SPEARS framework — scope, principles, examples, assertions, refinements, with ReAct and multi-agent patterns