dockeeper
Pass
Audited by Gen Agent Trust Hub on Apr 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes
git logandpython3to analyze the repository's commit history and parse local task data in.team/task-state.json.\n- [PROMPT_INJECTION]: The skill represents an indirect prompt injection surface as it ingests content from git commit messages and task description files (.team/tasks/*.md) to determine documentation updates.\n - Ingestion points: Data is pulled from git logs and markdown files in the
.team/tasks/directory.\n - Boundary markers: The skill does not define clear boundaries or instructions for the agent to ignore potentially malicious directions within the processed files.\n
- Capability inventory: The agent has the ability to write to documentation files in the
docs/spec/folder based on its interpretation of the ingested data.\n - Sanitization: There is no evidence of sanitization or filtering applied to text retrieved from commit logs or task files.
Audit Metadata