research-design-helper
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill documentation (scripts/README.md) instructs the user to install the 'docx' package from npm, which is a well-known registry for Node.js packages. This is a standard and expected dependency for the skill's document conversion functionality.
- [COMMAND_EXECUTION]: The agent is instructed in SKILL.md to execute a local Node.js script (scripts/brief_to_docx.js) to convert research design briefs from Markdown to Word format. The script performs local file operations and does not exhibit malicious behavior.
- [INDIRECT_PROMPT_INJECTION]: The skill defines a surface for indirect prompt injection by reading data from external files to pre-fill the research brief. However, the requirement for a Socratic dialogue with the user provides a layer of human review.
- Ingestion points: .research/project_manifest.yml, .research/topic_dossier.gaps.yml, .research/design_brief.md, and .research/literature_matrix.md (as specified in the SKILL.md 'Inputs' section).
- Boundary markers: The skill uses 'TODO' placeholders in its template (references/design_brief_template.md) to indicate sections to be filled but does not implement formal boundary markers for interpolated data.
- Capability inventory: The skill has capabilities to read from and write to the local filesystem and execute a local conversion script (scripts/brief_to_docx.js).
- Sanitization: No explicit sanitization or escaping of the ingested data is defined in the instructions before it is interpolated into the Markdown brief.
Audit Metadata