niar-web-presentation
Pass
Audited by Gen Agent Trust Hub on May 11, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to aggregate and synthesize information from multiple external documents as outlined in
multi-document-workflow.md. This ingestion process presents a potential surface for indirect prompt injection if input documents contain malicious instructions intended to manipulate the agent's summary or the generated presentation content. - Ingestion points: Multi-file workflows and project directory reading.
- Boundary markers: No explicit delimiters or instructions to ignore embedded commands are mandated for the ingestion phase.
- Capability inventory: The skill can write files to the local file system and generate HTML files containing JavaScript.
- Sanitization: The
scaffold-niar-web-presentation.pyscript uses proper HTML escaping and JSON serialization to sanitize inputs during the generation of theindex.htmltemplate. - [DYNAMIC_EXECUTION]: The Python scaffolding script programmatically generates an
index.htmlfile that includes a script block for presentation logic. This is a functional requirement of the skill and uses a static, well-defined template for code generation. - [DATA_EXPOSURE]: Documentation and examples in
READEM.mdandusage-examples.mdreference absolute file paths on a local drive (e.g.,D:\antigravity\...). These appear to be illustrative examples for the intended user environment and do not involve unauthorized access to sensitive system directories.
Audit Metadata