The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes Node.js and Shell scripts (archive-change.mjs, check-change-record.mjs, install-archivist-hooks.sh) that execute local git commands to gather metadata about repository changes. It also provides a method to install Git hooks in the .git/hooks directory to automate documentation checks.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by ingesting and summarizing Git diffs and status outputs which could contain untrusted content. This risk is mitigated by explicit instructions to wrap such content in markdown fences and to strictly avoid recording sensitive values like credentials or private tokens.
Ingestion points: Git status and diff outputs are read into the agent context via archive-change.mjs.
Boundary markers: Templates in archive-change.mjs use markdown code fences to delimit untrusted Git output.
Capability inventory: The skill uses fs.writeFileSync for file updates and spawnSync for Git operations.
Sanitization: Content is escaped into markdown blocks to prevent accidental instruction execution.
[SAFE]: All identified behaviors, including file system modifications and Git hook installation, are aligned with the skill's documented purpose of repository archival and history maintenance. No evidence of obfuscation or data exfiltration was found.

archivist