The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes conversation history to generate a handoff document, which creates a surface for instructions embedded in previous conversation data to influence the agent's output.
Ingestion points: The skill ingests the entire conversation history and optional user-provided arguments.
Boundary markers: No specific delimiters or boundary markers are defined for the output document to separate summarized content from agent instructions.
Capability inventory: The skill utilizes file-writing capabilities to save the resulting document to the filesystem.
Sanitization: The instructions explicitly mandate the redaction of sensitive information, including API keys, passwords, and PII, which serves as a security control for data exposure.
[COMMAND_EXECUTION]: The skill instructs the agent to perform file system operations outside of the active project workspace.
Evidence: "Save to the temporary directory of the user's OS
not the current workspace." While the temporary directory is a standard location for transient files, writing outside the workspace is a capability that should be monitored for unexpected behavior.

handoff