workspace-builder
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill architecture describes a Retrieval-Augmented Generation (RAG) system using Qdrant. This creates an inherent attack surface where data ingested into the vector store could potentially contain malicious instructions intended to manipulate agent responses.
- Ingestion points: Data indexed within the Qdrant vector database (SKILL.md).
- Boundary markers: Not explicitly defined in this architectural overview.
- Capability inventory: The system design includes execution of shell scripts and interaction with MCP servers.
- Sanitization: The specification does not currently detail input validation or sanitization for retrieved content.
- [COMMAND_EXECUTION]: The documentation provides templates and examples for initializing projects using shell commands.
- Evidence: Examples including git clone, ./scripts/setup.sh, and ./scripts/init-project.sh are provided for workspace configuration.
Audit Metadata