The Agent Skills Directory

[COMMAND_EXECUTION]: The skill implements a self-evolution mechanism (§UTE) that performs file system operations. It records usage data to the .skill-audit/ directory and possesses the capability to apply micro-patches to its own source file (skill.md), including updating triggers and versioning. This autonomous self-modification of the skill's instruction set represents a dynamic execution and persistence pattern.
[PROMPT_INJECTION]: The skill processes external data from API responses and user-provided test suites. While it includes security checks (CWE-89/78), the interpolation of untrusted third-party data into the agent's context provides a surface for indirect prompt injection. 1. Ingestion points: API response bodies in TEST/BATCH modes; test case arrays in BATCH mode. 2. Boundary markers: The skill instructs the agent to sanitize inputs (CWE-89) and mask PII (CWE-200), though no explicit delimiters for raw response data are defined in the output templates. 3. Capability inventory: Performs HTTP requests with multiple methods; writes to the file system for auditing and self-patching. 4. Sanitization: Employs masking and sanitization protocols based on CWE standards as described in §8.

api-tester