The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read and follow instructions found in local project files such as CONTEXT.md and DESIGN.md. This creates a data ingestion surface where malicious instructions placed in those files by an attacker could potentially influence the agent's behavior during the development process.
Ingestion points: The instructions in SKILL.md direct the agent to read CONTEXT.md for domain vocabulary and DESIGN.md for mandatory behavioral states.
Boundary markers: The skill does not provide any specific boundary markers or instructions to treat the content of these files as untrusted data.
Capability inventory: The skill influences the agent's code-writing and file-system manipulation capabilities within the user's workspace.
Sanitization: There are no mentions of sanitizing or validating the contents of the referenced files before they are processed by the agent.

tdd