record
Pass
Audited by Gen Agent Trust Hub on Jul 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes Git commands including
git rev-parse,git check-ignore, andgit ls-filesto manage file visibility and ensure the record file is not included in version control history. - [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it reads and reconciles state from an external file (
RECORD.md). - Ingestion points: Reads from
RECORD.mdat the repository root or a user-specified path. - Boundary markers: The record uses a standard markdown template but lacks specific security delimiters to segregate untrusted content.
- Capability inventory: The skill can read/write to the file system and execute Git commands via the shell.
- Sanitization: The instructions include a mitigation strategy requiring the agent to represent sensitive values by their purpose rather than their literal content.
Audit Metadata