Commands
| Command | Usage | Detail |
|---|---|---|
| init | opengap init --template <minimal|standard|full|llm-wiki> | Templates: minimal (2 files), standard (skills + tools), full (compliance + hooks + memory), llm-wiki (knowledge base) |
| validate | opengap validate --compliance | JSON schema validation, skill checks, and optional regulatory compliance audit |
| run | opengap run -a <adapter> -p "prompt" | Adapters: claude, openai, crewai, openclaw, nanobot, lyzr, github, opencode, gemini, gitclaw, git, prompt |
| export | opengap export --format <format> -o output | Formats: system-prompt, claude-code, openai, crewai, openclaw, nanobot, lyzr, github, copilot, opencode, cursor, gemini, codex, kiro, gitclaw |
| import | opengap import --from <format> <path> | Formats: claude, cursor, crewai, opencode, gemini, codex |
| install | opengap install | Shallow-clones dependencies at specified versions into mount paths |
| skills | opengap skills search "code review" | Registries: SkillsMP marketplace, GitHub repos, local filesystem |
| audit | opengap audit | FINRA 3110, SEC 17a-4, SR 11-7, CFPB checks with pass/fail/warn indicators |
| info | opengap info | Shows config, model, skills, tools, compliance, and SOUL.md preview |
| lyzr | opengap lyzr run -r <repo> -p "Hello" | Subcommands: run (clone → create → chat), create (register agent), update (push changes), list (show all agents). Saves agent ID for reuse |
opengap run Flags
opengap run -r "https://github.com/org/agent-repo" -a claude
| Flag | Description |
|---|---|
-r <url> | Remote repo URL to clone and run |
-a <adapter> | Adapter to use (claude, openai, github, gemini, etc.) |
-p "prompt" | One-shot prompt (non-interactive) |
-b <branch> | Run a specific git branch |
-d <dir> | Agent directory (local) |
-w <dir> | Working directory for the spawned agent process |
--refresh | Force re-clone even if cached |
--no-cache | Clone to temp dir, delete on exit |
opengap export Formats
opengap export --format claude-code -o ./output
| Format | Output |
|---|---|
claude-code | CLAUDE.md with skills, model hints, and compliance |
openai | Python code with Agent(), tool stubs, and type mappings |
crewai | YAML crew config with role/goal extraction |
system-prompt | Single concatenated Markdown for any LLM |
cursor | Cursor rules file with agent identity and skills |
gemini | Gemini system instruction and tool definitions |
lyzr | API payload with provider mapping and credential IDs |
copilot | GitHub Copilot workspace instructions |
opencode | OpenCode system prompt and tool configuration |
codex | OpenAI Codex agent config with tool stubs |
kiro | Kiro agent spec with steering and hook definitions |
openclaw | Workspace with config JSON, AGENTS.md, tools, and skills |
nanobot | Config JSON + system prompt for Nanobot runtime |
gitclaw | GitClaw workspace with config and skill mapping |
Adapters
Full adapter table: modes, requirements, features
Compliance
opengap audit: FINRA, SEC, Federal Reserve checks
Quick Start
Init, validate, and run your first agent
Skills CLI
opengap skills search, install, list, info