The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute several shell commands for status checks, testing, and communication, including rig whoami --json, npm test, and rig send. These are standard tools within the expected execution environment.
[DYNAMIC_EXECUTION]: The instructions encourage the agent to verify findings by writing and executing reproduction scripts using npx tsx -e "...". This allows for the runtime generation and execution of code based on the agent's analysis of the codebase.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves ingesting and processing untrusted data from external sources during the review process.
Ingestion points: The agent reads project source code, configuration files (e.g., project.yaml, mission.yaml), and documentation (e.g., CLAUDE.md, architecture docs) as part of its review task.
Boundary markers: No explicit instructions or delimiters are provided to the agent to distinguish between its own instructions and potentially malicious instructions embedded in the code or comments it is reviewing.
Capability inventory: The agent has access to shell execution (npm, npx), communication tools (rig send, rig chatroom), and file system write access for saving review artifacts to the docs/review/ directory.
Sanitization: No sanitization or filtering of the ingested external content is mentioned before the agent processes it or incorporates it into its verification logic.

review-team