The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute mktemp to create a temporary file and write the conversation summary to it. This involves direct shell command invocation to manage file paths.
[DATA_EXFILTRATION]: Writing conversation summaries to the /tmp directory (as generated by mktemp) can expose sensitive information to other local users on a multi-user system, as these directories are often globally readable.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by processing untrusted data from the conversation history.
Ingestion points: Conversation history and user-supplied focus description in SKILL.md.
Boundary markers: None present; the agent is not instructed to use delimiters or ignore embedded commands within the conversation data being summarized.
Capability inventory: File system write access via the summarized content output.
Sanitization: No explicit sanitization or validation of the conversation content is performed before it is included in the handoff document.

handoff