The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes external data which creates an attack surface for indirect prompt injection.
Ingestion points: SKILL.md specifies that the agent should inspect "files, diffs, logs, docs, tests, browser state, or prior messages" to gather context for teaching.
Boundary markers: There are no explicit markers or instructions to treat this content as untrusted or to ignore embedded instructions.
Capability inventory: The skill performs local file writes (.agents/teach-back.md) and can execute commands such as "run tests" or "guide debugger inspection".
Sanitization: No sanitization or filtering of the source material is mentioned before processing.
[COMMAND_EXECUTION]: The skill is instructed to perform local execution of tests and debugging tools.
Evidence: SKILL.md guides the agent to "run tests, or guide debugger inspection" when it helps the user's understanding.
[PROMPT_INJECTION]: The skill reads potentially sensitive local information to establish session context, providing a data source for the injection surface.
Evidence: SKILL.md instructs the agent to inspect files, logs, and browser state before beginning the teaching loop.

teach-back