research-prompt
Pass
Audited by Gen Agent Trust Hub on Jul 7, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a process for ingesting external data (project files and conversation history) to generate a research prompt. This creates a surface for indirect prompt injection.
- Ingestion points: SKILL.md (Process Step 1) instructs the agent to pull context from relevant project files and conversation history.
- Boundary markers: Absent; the template does not use specific delimiters or instructions to ignore nested commands within the ingested context.
- Capability inventory: The skill documentation mentions executing the resulting prompt via an external API endpoint (
DeepAPI). - Sanitization: No explicit sanitization or escaping of the ingested data is defined before it is interpolated into the final prompt.
- [DATA_EXPOSURE]: The skill's primary function involves reading project files and conversation history to build context. Users should ensure that files accessible to the agent do not contain sensitive credentials, as the skill aims to summarize this information for a research task.
- [EXTERNAL_DOWNLOADS]: The skill mentions an external network endpoint (
DeepAPI POST /v1/research/deep) for execution. This is presented as an instructional workflow rather than an automated script execution within the skill itself.
Audit Metadata