drafting-slides
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a local Node.js script (scripts/drafting-slides.mjs) to generate a visuals plan. The agent is specifically directed to avoid shell-based injection vectors like command chaining or heredocs when running this command.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external files (storyline.md, brief.md, sources.md) which serve as an attack surface. * Ingestion points: Content is pulled from user-controlled files in the deck folder. * Boundary markers: The workflow relies on an 'approved: true' metadata field to ensure manual review of inputs. * Capability inventory: The skill can perform file operations and execute local Node.js scripts. * Sanitization: There is no explicit sanitization logic; the skill relies on the agent's prose generation and the optional 'unslop' tool.
- [DYNAMIC_EXECUTION]: The skill uses a provided script that imports modules locally (scripts/deck.mjs) to parse Markdown content at runtime.
Audit Metadata