brainstorming
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the agent to execute a Python-based utility script named
<aegis-workspace-helper>with various subcommands (init, append-index, check) to manage the local documentation workspace. - [INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface for indirect prompt injection by processing external project data. \n
- Ingestion points: As defined in
SKILL.md, the agent is instructed to explore project context, read authority docs, and investigate the codebase. \n - Boundary markers: The skill lacks explicit markers or instructions to the agent to disregard instructions that may be embedded within project files. \n
- Capability inventory: The agent has the authority to write new documentation files to the disk and execute local scripts. \n
- Sanitization: There is no process described for sanitizing or validating the content read from project files before it influences agent behavior.
Audit Metadata