The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill reads existing project documentation which represents an attack surface for indirect prompt injection if those files contain malicious instructions.
Ingestion points: The skill instructions in SKILL.md mandate reading docs/adr/ and README.md before performing any actions.
Boundary markers: There are no specific instructions or delimiters defined to isolate content read from files or to warn the agent to ignore embedded instructions within those files.
Capability inventory: The skill is designed to update README.md and create/modify files in docs/adr/, providing a persistent write capability to the local filesystem.
Sanitization: The instructions do not specify any sanitization, filtering, or validation of the content extracted from the existing documentation before it is processed or used to generate new documentation.

brain-to-docs