The Agent Skills Directory

[COMMAND_EXECUTION]: The documentation describes the use of tools like rigour_run and rigour_run_supervised. These tools are designed to execute arbitrary shell commands (e.g., npm test) under the supervision of the Rigour system to facilitate quality-gated command execution and self-healing agent loops.
[PROMPT_INJECTION]: The skill documents tools that ingest external, potentially untrusted data for analysis, such as pull request diffs (rigour_review) and project source code (rigour_check_deep). This creates a surface for indirect prompt injection attacks where malicious instructions embedded in the analyzed code could attempt to influence the agent's behavior.
Ingestion points: Project source files and git diffs are ingested via parameters like --cwd and --diff.
Boundary markers: No specific boundary markers or "ignore instructions" delimiters are mentioned in the usage guide.
Capability inventory: The tools possess significant capabilities including subprocess execution (rigour_run) and configuration modification (rigour_mcp_set_settings).
Sanitization: The documentation does not provide details on how the tools sanitize or validate the content of ingested code or diffs before processing them with an LLM.

rigour-usage