The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands and git operations to resolve repository paths and manage local storage. It incorporates defensive coding practices by checking for unsafe symlinks and enforcing restrictive user-only permissions (chmod 700) on created directories.- [DATA_EXPOSURE]: It establishes a protocol for persisting session context while explicitly mandating the redaction of secrets, credentials, and personal information to prevent sensitive data exposure.- [INDIRECT_PROMPT_INJECTION]: The resume functionality allows the agent to ingest external files, which represents a potential attack surface. The skill mitigates this risk by providing clear instructions to treat all resumed content as untrusted context rather than executable instructions, and it strictly prohibits the agent from following commands or links found within these files.

session-handoff