Quick Start
Run inside any repository to launch the interactive 2-step TUI:
npx skillops-cli@latest add
| 1. Select Your AI IDE / Agent | 2. Select Skills & Subagents |
|---|
| |
CLI Reference
Install globally (npm i -g skillops-cli) or run directly with npx skillops-cli@latest:
| Command | Action |
|---|
skillops add | Open the interactive TUI (select target AI IDEs → select Skills & Subagents) |
skillops add <name...> -t cursor,claude | Non-interactive install into specific targets (-t all for all 28 tools) |
skillops status (list, ls) | Inspect installed Skills, Subagents, and SHA-256 health across AI harnesses |
skillops update [name...] (sync, upgrade) | Refresh installed modules to the latest catalog version while preserving local edits |
skillops remove [name...] | Interactively or directly uninstall managed artifacts and update .skillops-lock.json |
Flags: --all selects every item in the catalog, --dry-run previews target file paths without writing to disk, and --force overwrites locally modified files.
Why SkillOps?
◆ 28-IDE Native Transpiler
Author once in catalog/. SkillOps compiles each SKILL.md (plus bundled scripts/ & references/) and SUBAGENT.md into the exact folder hierarchy and frontmatter required by each IDE or CLI.
⚡ 3-Tier Token Budget
Prevents context-window bloat: L1 YAML routing (≤ 160 tokens), L2 workflow instructions (≤ 250 lines), and L3 deterministic Python tools (0 idle tokens). Subagents run in isolated child contexts (≤ 400 output tokens).
🔒 SHA-256 Drift Protection
Every generated file is tracked in .skillops-lock.json with a cryptographic digest. Manual edits in .cursor/, .claude/, or .github/ are never overwritten or deleted unless --force is passed.
Supported AI IDEs & CLIs (28)
| Target ID | AI IDE / CLI | Skills Output Path | Subagents Output Path |
|---|
cursor | Cursor AI IDE | .cursor/rules/skillops-<name>.mdc | .cursor/rules/skillops-agent-<name>.mdc |
claude | Anthropic Claude Code | .claude/skills/<name>/SKILL.md | .claude/agents/<name>.md |
copilot | GitHub Copilot | .github/instructions/skillops-<name>.instructions.md | .github/agents/<name>.agent.md |
antigravity | Google Antigravity IDE | .agent/skills/<name>/SKILL.md | .agent/agents/<name>.md |
windsurf | Windsurf AI IDE | .windsurf/rules/skillops-<name>.md | .windsurf/workflows/skillops-agent-<name>.md |
codex | OpenAI Codex CLI | .codex/skills/<name>/SKILL.md + AGENTS.md | .codex/agents/<name>.md |
gemini | Google Gemini CLI | .gemini/skills/<name>/SKILL.md | .gemini/agents/<name>.md |
cline | Cline | .clinerules/skillops-<name>.md | .clinerules/workflows/skillops-agent-<name>.md |
roocode | |
| Target ID | AI IDE / CLI | Skills Output Path | Subagents Output Path |
|---|
kilocode | Kilo Code | .kilocode/rules/skillops-<name>.md | .kilocode/workflows/skillops-agent-<name>.md |
zed | Zed AI | .zed/rules/skillops-<name>.md | .zed/rules/skillops-agent-<name>.md |
continue | Continue.dev | .continue/rules/skillops-<name>.md | .continue/prompts/skillops-agent-<name>.prompt |
aider | Aider | .aider/conventions/skillops-<name>.md | .aider/conventions/skillops-agent-<name>.md |
trae | Trae AI IDE | .trae/rules/skillops-<name>.md | .trae/agents/<name>.md |
augment | Augment Code | .augment/rules/skillops-<name>.md | .augment/agents/<name>.md |
factory | Factory AI Droid | .factory/skills/<name>/SKILL.md | .factory/droids/<name>.md |
qwen | Qwen Code CLI | .qwen/skills/<name>/SKILL.md | .qwen/agents/<name>.md |
opencode | OpenCode CLI | |
Security & Trust Model (5-Pillar Defense-in-Depth)
Installing AI Skills and executable helper scripts into your IDE requires strict supply-chain guarantees. SkillOps enforces a 5-Pillar Security Architecture (.github/SECURITY.md & INV-SKILL-06 in SKILL_CONTRACT.md) so no unreviewed or malicious skill can enter the catalog:
This is the beginning of the README. Read the rest at the original. Read the full README ↗
