The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the execution of local shell scripts (e.g., .guardrails/scripts/find-items.sh, new-id.sh, check-trace.sh) to manage the risk management ledger. This is expected behavior for an integrated development tool.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from existing risk items and requirements within the repository, which represents a potential attack surface for indirect prompt injection if the repository content is untrusted.
Ingestion points: Data is read from the risk management ledger and individual items via the find-items.sh show command.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to disregard potential instructions embedded within the ingested risk data.
Capability inventory: The agent has the capability to execute local scripts and create/modify files in the docs/risk/ directory.
Sanitization: No explicit sanitization or escaping is mentioned for the data read from the repository, though the output must conform to a specific grammar parsed by check-trace.sh.

analyze-risks