orchestrate-batch-refactor
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes repository source code as its primary input, which creates a potential surface for indirect prompt injection if the code being refactored contains malicious instructions designed to influence the agent's behavior.
- Ingestion points: Target repository paths and source code modules analyzed by explorer agents as defined in SKILL.md.
- Boundary markers: The skill defines strict ownership boundaries for worker agents and instructs them to ignore unrelated edits, providing isolation between tasks (SKILL.md, agent-prompt-templates.md).
- Capability inventory: The skill facilitates file modifications and the execution of local validation commands (e.g., npm run test) through sub-agents (SKILL.md, work-packet-template.md).
- Sanitization: The workflow includes a synthesis phase and integration gates to review packet outcomes and resolve conflicts, acting as a manual and automated verification step (SKILL.md).
- [COMMAND_EXECUTION]: The skill utilizes local shell commands for the purpose of validating refactors (e.g., running test suites), which is an intended and documented feature of the skill's validation strategy.
Audit Metadata