The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing user-controlled inputs within its workflow. Ingestion points: The 'target', 'threat', and 'constraints' parameters in the input table. Boundary markers: The skill does not define delimiters or provide instructions to the agent to treat these inputs as untrusted data. Capability inventory: Access to 'Bash', 'Write', 'Edit', 'Read', 'Grep', and 'Glob' tools. Sanitization: There is no requirement or evidence of sanitization for the 'target' variable before it is interpolated into shell commands. \n- [COMMAND_EXECUTION]: The workflow relies on local command execution to perform security audits and verify mitigations. Evidence: Employs 'Bash' and 'Grep' to inspect application code for authentication checks and database migrations for Row Level Security (RLS) policies, as seen in the Workflow and Release Gate sections.

security-hardening