菜单
Scaffold and build projects for the lythoskill ecosystem only. Creates thin-skill monorepos: heavy logic in npm packages (Starter), agent-facing instructions in lightweight SKILL.md (Skill), build output committed to skills/ (Output).
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Weekly synthesis writer. Distills the past 7 days' core thread + quest DAG into a frontmatter-rich short doc. Never replays git log or cortex INDEX (those are direct queries already). Forms the weekly counterpart to project-scribe (daily) and project-onboarding (read-side).
Weekly synthesis writer. Distills the past 7 days' core thread + quest DAG into a frontmatter-rich short doc. Never replays git log or cortex INDEX (those are direct queries already). Forms the weekly counterpart to project-scribe (daily) and project-onboarding (read-side).
Session context loader. Reads the latest daily handoff to restore project context without redundant file exploration. Three-layer loading: CLAUDE.md (stable) → daily handoff (session state) → git verification (ground truth). Degrades to file exploration when no handoff exists. CQRS read-side pair with project-scribe.
Analyzes SKILL.md files against Agent Skills best practices. Reviews body size, description quality, progressive disclosure, frontmatter usage, and context efficiency. Provides actionable optimization advice.
夜有所梦 — project memory consolidation. Daily (scribe) captures raw experience, weekly extracts patterns and anomalies, dreaming consolidates the accumulated memory into SSOT. External review via ZK subagent (the same de facto standard pattern used by Hermes Curator's forked-agent review), with cross-model validation via arena for critical docs.
Session context loader. Reads the latest daily handoff to restore project context without redundant file exploration. Three-layer loading: CLAUDE.md (stable) → daily handoff (session state) → git verification (ground truth). Degrades to file exploration when no handoff exists. CQRS read-side pair with project-scribe.
| name | lythoskill-creator |
| version | 0.17.2 |
| description | Scaffold and build projects for the lythoskill ecosystem only. Creates thin-skill monorepos: heavy logic in npm packages (Starter), agent-facing instructions in lightweight SKILL.md (Skill), build output committed to skills/ (Output). |
| when_to_use | Scaffold a lythoskill monorepo, init a lythoskill project, add a skill to an existing lythoskill repo, build a skill that follows the thin-skill pattern (Starter + Skill + Output), lythoskill skill template, new lythoskill repository. NOT for: writing a generic skill (just create SKILL.md directly), creating a plain npm CLI tool (use your usual scaffolding). |
Not a generic skill creator. This scaffolds projects for the lythoskill ecosystem only. If you just want to write a skill, create a
SKILL.mdfile directly — no tooling needed.
A lythoskill project separates concerns into three layers:
packages/<name>/): npm-publishable package with CLI. All dependencies managed here.packages/<name>/skill/): Thin SKILL.md + scripts that call the starter via bunx.skills/<name>/): Build output copied from Skill layer. Committed to Git so consumers can clone and use without building.Scaffold a new lythoskill monorepo.
bunx @lythos/skill-creator@0.17.2 init <project-name>
Creates a monorepo with a starter package and an example skill.
Add a new skill to an existing lythoskill monorepo. Follows the cortex Step 1/2/3 pattern: CLI creates template → agent fills content → probe verifies.
bunx @lythos/skill-creator@0.17.2 add-skill <skill-name>
Step 1 (CLI): Creates packages/<skill-name>/skill/SKILL.md template.
Step 2 (agent): Edit SKILL.md — fill description, when_to_use, body, gotchas.
→ After editing, run bunx @lythos/skill-coach to validate against 12-dimension quality rubric.
→ Fix issues coach flags, then proceed to Step 3.
Step 3 (verify): Run cortex probe to confirm no empty-shell template.
Two skill types, decided upfront:
| Type | Has CLI? | Has package.json? | npm publish? | Example |
|---|---|---|---|---|
| Skill + CLI | ✅ bin field | ✅ required | ✅ to npm | deck, arena, curator, cortex |
| Pure skill | ❌ no CLI | ❌ none needed | ❌ skip | scribe, onboarding, coach, journalist |
add-skill creates packages/<name>/ with package.json + src/cli.ts + skill/SKILL.md. Add to scripts/publish.sh PACKAGES array.packages/<name>/skill/SKILL.md directly. No package.json, no CLI, no npm. Distribution via cold pool (git clone / localhost). The pre-commit build hook auto-creates skills/<name>/ output.After creation: run bunx @lythos/skill-creator build <name> to build the skill output, then git add skills/<name>/.
Build a skill for distribution. Copies packages/<name>/skill/ to skills/<name>/, strips dev-only files, validates SKILL.md frontmatter, and enforces the unified version.
bunx @lythos/skill-creator@0.17.2 build <skill-name>
Build all skills at once (used by pre-commit):
bunx @lythos/skill-creator@0.17.2 build --all
Audit an existing project against current lythoskill conventions. Reports drift between your project and the latest scaffolding standards.
# Audit only — reports what would change
bunx @lythos/skill-creator@0.17.2 align
# Auto-fix drift where possible
bunx @lythos/skill-creator@0.17.2 align --fix
Checks for: outdated package.json scripts, stale .gitignore patterns, missing skill-layer boilerplate, and version mismatches.
Lock-step version bump for the entire monorepo. Updates root package.json + every packages/*/package.json to the same version (per project_lockstep_versioning policy). Used before npm publish runs.
# Semver targets
bunx @lythos/skill-creator@0.17.2 bump patch
bunx @lythos/skill-creator@0.17.2 bump minor
bunx @lythos/skill-creator@0.17.2 bump major
# Explicit version (e.g. for major milestones)
bunx @lythos/skill-creator@0.17.2 bump 1.0.0
# Preview the bump without writing files
bunx @lythos/skill-creator@0.17.2 bump 1.0.0 --dry-run
Run from the project root. Expects source at packages/<name>/skill/ and outputs to skills/<name>/.