The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it ingests untrusted data from the project being audited (code, documentation, and metadata) to generate reports and create project issues. \n
Ingestion points: The skill reads all project files, documentation in SUMMARY.md, and feature records via bd list. \n
Boundary markers: The instructions lack explicit delimiters or guidance for the agent to ignore instructions embedded within the audited files. \n
Capability inventory: The skill can execute shell commands, create project issues, and push to a remote database. \n
Sanitization: No sanitization is performed on the ingested data before it is used to populate audit reports or issue descriptions. \n- [COMMAND_EXECUTION]: The skill executes local tools including the bd (Beads) CLI and a project-specific Python script scripts/check-docs.py via uv run. These are used for project reconciliation and documentation validation. \n- [DATA_EXFILTRATION]: The skill performs authorized synchronization of project state to a remote Dolt database using bd dolt push.

audit-project