llm-wiki
Pass
Audited by Gen Agent Trust Hub on Apr 29, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an Indirect Prompt Injection surface during its ingestion and processing phases.
- Ingestion points: The skill reads and processes external data from the raw/ directory (SKILL.md), including web content, YouTube transcripts, and documents fetched from the web (Phases 8 and 9).
- Boundary markers: The skill lacks instructions to use delimiters or specific warnings to ignore instructions embedded within the processed raw data.
- Capability inventory: The agent has access to Read, Write, and Bash tools (SKILL.md) which are used during the acquisition and ingestion phases.
- Sanitization: There is no evidence of sanitizing or validating the content of external sources before they are processed by the agent.
- [COMMAND_EXECUTION]: The skill provides bash snippets in the preamble and inventory steps (Phases 8 and 9) to perform git branch checks and file discovery.
Audit Metadata