The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to search through memory/history.jsonl using the grep tool. Since this file contains raw historical conversation data, it may include malicious instructions from previous sessions that could influence the agent's current behavior if retrieved.
Ingestion points: memory/history.jsonl (referenced via the system prompt's history log path).
Boundary markers: None identified. The skill does not provide delimiters or instructions to treat the search results as untrusted data.
Capability inventory: The skill utilizes the grep tool to read file contents.
Sanitization: No sanitization, filtering, or validation steps are described for the data retrieved from the history logs.
[DATA_EXPOSURE]: The skill documentation explicitly provides a search example for sensitive patterns (oauth|token). This confirms the capability and intent to retrieve potentially sensitive authentication credentials from historical conversation logs.

memory