The Agent Skills Directory

[SAFE]: The skill instructions and associated rules were analyzed for prompt injection patterns. All instructions are focused on guiding the agent's logic for TDD and do not attempt to bypass safety filters or override ethical guidelines.
[SAFE]: Example source code and fixtures in the evals/ directory were inspected for data exfiltration and credential exposure. The examples use documentation-reserved domains (e.g., pay.example) and do not contain hardcoded secrets or access to sensitive local paths.
[SAFE]: No remote code execution or unverifiable dependency patterns were found. The skill does not perform external downloads or command-line executions; it relies entirely on the provided documentation and example files.
[SAFE]: Analysis for obfuscation, including Base64-encoded strings, zero-width characters, and homoglyphs, returned no findings. The Base64 usage in example code is standard implementation logic for JWT decoding and not a concealment technique.
[SAFE]: The skill does not implement any persistence mechanisms or privilege escalation techniques. It operates as a set of logical instructions for software development workflows.

test-first-by-evidence