The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were identified. The skill instructions prioritize security and reliability through several mechanisms:
[INDIRECT_PROMPT_INJECTION]: The skill functions as an orchestration layer processing user-provided task descriptions via the $ARGUMENTS variable in SKILL.md. While no explicit boundary markers or sanitization are defined for the subagent prompts, the risk is significantly mitigated by the 'Step 4' instruction to manually review every diff and output before synthesis, ensuring the orchestrator does not blindly trust subagent actions.
[COMMAND_EXECUTION]: The skill instructs the agent to execute verification commands (tests, linting, or builds) discovered within the target repository. This is an expected part of the software development lifecycle and is executed in a controlled context where the orchestrator manages the final state.
[PRIVILEGE_ESCALATION]: The skill enforces a least-privilege model for subagents by explicitly prohibiting them from performing git operations (commits or pushes), reserving those actions solely for the main orchestrator agent after verification.

orchestrate