vectorcode
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill documentation provides examples for using the 'vectorcode' CLI tool to initialize projects, index source directories, and perform manual queries (e.g., 'vectorcode init', 'vectorcode vectorise src/').
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data by indexing local source code files for retrieval during semantic search.
- Ingestion points: The 'mcp__vectorcode__vectorise' tool and the 'vectorcode update' command read source code files from the local filesystem (SKILL.md).
- Boundary markers: The instructions do not define specific delimiters or instructions to ignore potential prompts embedded within the indexed code comments or strings.
- Capability inventory: The skill includes tools to list indexed projects, query the vector database, and manage indexed files ('ls', 'query', 'vectorise', 'files_ls', 'files_rm').
- Sanitization: There is no mention of content filtering or sanitization performed on the file contents before they are added to the vector index or returned as search results.
Audit Metadata