The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to read external project files to align testing vocabulary and domain logic, creating a surface for potential indirect prompt injection.
Ingestion points: CONTEXT.md is explicitly referenced in SKILL.md as a source for domain language and vocabulary.
Boundary markers: The skill does not provide explicit boundary markers or instructions for the agent to ignore potentially malicious directives embedded within the ingested context.
Capability inventory: The skill is intended for use by coding agents with the ability to perform codebase exploration, call other skills (codebase-design, code-review), and write/execute tests, implying significant file system and shell interaction capabilities.
Sanitization: There are no instructions or mechanisms provided for sanitizing or validating the content of external documentation files before they are processed by the agent.

tdd