The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified. The skill is designed as a local-first utility for task state management.
[INDIRECT_PROMPT_INJECTION]: The skill processes task context from previous sessions. It mitigates injection risks by wrapping data in boundary markers (<brb-checkpoint-data>), including warnings to the agent, and scanning content for suspicious patterns such as instruction overrides or destructive commands.
[DATA_EXFILTRATION]: The skill operates entirely on the local file system within the user's home directory. It includes comprehensive logic to redact secrets from task context and excludes sensitive file types (like environment files and private keys) from repository snapshots, preventing accidental data exposure.
[COMMAND_EXECUTION]: The skill interacts with the local git repository using secure methods (execFileSync) that are resistant to shell injection. Commands are limited to standard git status and metadata operations.

brb