ask-me-bro
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill relies on external content from
docs/guide/process/on-ramps.mdto decide which task to perform next. While this is a common architectural pattern for routing, it creates a surface where the agent's logic depends on untrusted file content. - Ingestion points: The documentation file
docs/guide/process/on-ramps.mdis accessed inSKILL.md. - Boundary markers: Absent. There are no explicit instructions for the agent to ignore instructions embedded within the documentation file.
- Capability inventory: Handing off or recommending specific development-focused skills and commands.
- Sanitization: Absent. No logic is present to filter or validate the content of the documentation file before it affects the agent's decision-making process.
- [SAFE]: The skill includes a
disable-model-invocation: truedirective in the frontmatter, which restricts the model from making unauthorized tool calls and forces a controlled routing flow.
Audit Metadata