The Agent Skills Directory

[PROMPT_INJECTION]: The skill retrieves records from memory tables which may include untrusted input from previous interactions.
Ingestion points: External data enters the context via the recall recent command in SKILL.md.
Boundary markers: The instructions do not define delimiters for the retrieved content or warn the agent to ignore embedded instructions.
Capability inventory: The skill facilitates reading memory; additional capabilities depend on the agent's environment.
Sanitization: There is no evidence of filtering or escaping for the data retrieved from the tables.

recall-recent