cli-vstash
Pass
Audited by Gen Agent Trust Hub on Apr 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes the
vstashCLI to manage document ingestion and search operations. - [EXTERNAL_DOWNLOADS]: The
vstash addcommand enables fetching content from remote web addresses. - [DATA_EXFILTRATION]: As described in the documentation, usage of cloud-based inference providers (OpenAI, Cerebras) involves sending document chunks to external endpoints.
- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface due to its RAG (Retrieval-Augmented Generation) workflow.
- Ingestion points: Data from external files and URLs is ingested via
vstash add(SKILL.md). - Boundary markers: No explicit markers or instructions are provided to the agent to treat ingested content as untrusted.
- Capability inventory: The skill includes an LLM-answering capability via
vstash ask. - Sanitization: No validation or sanitization of external content is specified before processing.
Audit Metadata