feature-presentation
Pass
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill synthesizes data from multiple external sources (initiatives, tasks, feature specs, and mockups) into a standalone HTML deck. This ingestion process is a known surface for indirect prompt injection, where malicious content in a project file could attempt to influence the agent's logic or compromise the generated artifact.
- Ingestion points: Artifacts are gathered from
{artifacts-root}and{spec-root}via globbing (referenced inSKILL.mdStep 1 andreferences/artifact-accumulation.md). - Boundary markers: The instructions mandate wrapping slide content in specific HTML structures and provide warnings about slide termination rules.
- Capability inventory: The skill has the ability to read/write to the filesystem, execute local scripts via
node, and spawn sub-agents for gap-filling tasks. - Sanitization: The skill contains robust instructions for HTML entity escaping (specifically prioritizing the
&character) to prevent cross-site scripting (XSS) or attribute breakout iniframe srcdocembeds (documented inreferences/deck-template.md§3). - [COMMAND_EXECUTION]: The skill relies on executing local Node.js scripts to perform conformance validation and file conversion.
- Evidence: Specifically invokes
node .claude/skills/presentation-builder/scripts/validate-presentation.cjsfor document verification andnode .claude/skills/html-export/scripts/export.cjsfor PDF generation.
Audit Metadata