| name | openclinxr-openclaw-style |
| description | Use when operating OpenClinXR's OpenClaw-style / OpenClaw-inspired repo-native workflow in Codex, including role consultation, run-next, leases, hooks, and canonical state updates. |
OpenClinXR OpenClaw-Style Bridge
This skill packages the repo-native OpenClaw-style / OpenClaw-inspired workflow for Codex. It is not an external OpenClaw runtime.
Start
- Read
AGENTS.md.
- Read the snapshot heads of
PROJECT_COORDINATION_INDEX.md, AUTONOMOUS_WORK_PLAN.md, and docs/openclinxr/worker-backlog-and-validation-matrix.md.
- Use
pnpm openclaw:run-next to select a real approved slice when continuation is needed.
- Acquire a lease before edits with
pnpm openclaw:lease -- acquire --owner <owner> --slice "<slice>" --ttl-minutes 60.
- Use Codex custom agents from
.codex/agents/*.toml only when explicitly asked for subagents/delegation or when the repo instructions require role consultation.
Guardrails
- Treat
agents/**/charter.md, agents/**/memory.md, and .agent-factory/memory-index.json as file-backed role memory.
- Keep canonical state in the repo files named by
AGENTS.md, not chat-only summaries.
- Run focused verification for touched code and
pnpm agent:alignment && pnpm docs:drift-check after coordination/OpenClaw-style changes.
