answer-me
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a locally provided JavaScript tool (
scripts/am.mjs) using the Node.js runtime. This execution is scoped to the skill's primary function of rendering visualization templates from Markdown content and does not involve downloading external binaries. - [INDIRECT_PROMPT_INJECTION]: The skill operates by ingesting user queries to generate structured explanations, creating a potential surface for indirect prompt injection.
- Ingestion points: User-provided requests for explanations, diagrams, or technical clarifications as described in the skill's trigger conditions (SKILL.md).
- Boundary markers: The instructions use a quoted shell heredoc (
'AM_EOF') to pass the generated content to the CLI tool, which effectively prevents shell parameter expansion and command injection at the boundary between the agent and the system shell (SKILL.md). - Capability inventory: The skill possesses the capability to execute a local Node.js script (
scripts/am.mjs) to process data and write it to an HTML file (SKILL.md). - Sanitization: The skill implements STE (Simplified Technical English) checks during the rendering process to enforce structural and linguistic constraints on the generated content, providing a layer of validation for the processed text (SKILL.md).
Audit Metadata