Security audit
Security checks for vulnerabilities and agentic risk
This is a clearly disclosed high-trust coding automation plugin; it can launch coding agents and land repository changes, but those powers fit its stated purpose and include review controls.
Install only if you trust this plugin to run coding agents against repositories the OpenClaw gateway user can access. Keep the default review-first settings unless you intentionally want more automation, and be cautious with full-access Codex settings, auto-merge/auto-PR modes, trusted verifier commands, and repositories with worktree setup scripts.
The description advertises broad natural-language triggers like "let oca do...", "ask oca to...", and "have oca handle...", which can overlap with ordinary conversation rather than a narrowly scoped command surface. In a high-privilege orchestration skill that launches and controls coding agents, accidental invocation could cause unintended session creation, delegation, or workflow actions.
The short-name trigger oca is described as a built-in alias and paired with vague examples, but the skill does not define strong scope constraints for when references to oca should actually invoke orchestration behavior. Because this skill can launch, resume, approve, merge, and escalate agent work, ambiguous triggering materially increases the chance of unintended privileged actions from benign mentions or conversational references.
No suspicious patterns detected.
