orchestrate
Pass
Audited by Gen Agent Trust Hub on Aug 24, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Node.js scripts to perform repository analysis and verify state changes.
- Evidence: Found in SKILL.md workflow documentation and the allowed-tools frontmatter field.
- [DYNAMIC_CONTEXT_INJECTION]: The skill utilizes dynamic context execution to automate the cleanup of its internal run metadata using a bundled script.
- Evidence: Found in SKILL.md under the 'FIFO' section: !node "${CLAUDE_PLUGIN_ROOT}/skills/orchestrate/scripts/prune-runs.js".
- [SAFE]: The skill implements advanced security practices, including object-identity verification (O_NOFOLLOW) to prevent symlink attacks and SHA-256 integrity bindings to ensure data consistency between process steps.
- Evidence: Detailed implementation and security rationale found across all bundled scripts in the scripts/ directory and references/admission-allowlist.json.
- [SAFE]: The skill processes untrusted repository data but incorporates robust mitigations against indirect prompt injection.
- Ingestion points: repo_signals (branch, head, features) in plan-context.js.
- Boundary markers: JSON schema enforcement and planner instructions in planner-prompt.md.
- Capability inventory: Explore agent with generic Bash capabilities.
- Sanitization: Output validation by validate-plan.js and mandatory human review (AskUserQuestion).
Audit Metadata