The Agent Skills Directory

Command Execution: The skill executes a local Python script (reference/scripts/advise.py) to generate security reports. This is a standard execution pattern for CLI-integrated tools and is documented within the command definition.
Indirect Prompt Injection Surface: The skill processes repository source code and database content, which are untrusted ingestion points. Ingestion points: The tool reads source files (e.g., src/auth.py) and a SQLite database (knowledge.db) into the agent's context. Boundary markers: The output is structured using markdown headers, though it lacks specific 'ignore instruction' delimiters for the ingested code content. Capability inventory: The skill has the capability to execute shell commands via the python3 CLI. Sanitization: The skill explicitly mitigates these risks in Protocol 2 by mandating the use of shlex.quote and absolute path resolution for all inputs, ensuring that data is handled according to secure implementation standards.
Secure Development Focus: The core logic of the skill is designed to prevent vulnerabilities by providing remediation patterns and threat model context, demonstrating a security-focused design that prioritizes defense-in-depth.

mantis-advise