The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions describe a 'Test-Driven Prompting (TDP)' pattern that captures verbatim test output (stdout and stderr) and interpolates it directly into the prompt for subsequent agent tasks. This creates a surface for indirect prompt injection where malicious code or tests could output instructions to manipulate the behavior of the implementer agent.\n
Ingestion points: Captured from execSync output during the TDD loop as described in SKILL.md and the 'Test-Driven Prompting (TDP)' section.\n
Boundary markers: The prompt template uses triple backticks (markdown code blocks) but lacks defensive instructions for the agent to treat the data as untrusted.\n
Capability inventory: The pattern uses the Task tool to spawn subagents and requires Bash or Edit capabilities for the implementation phase.\n
Sanitization: No sanitization or character escaping of the captured test output is implemented.

tdd