knowledge-base-builder
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The provided Bash CLI template contains a SQL injection vulnerability. In the
search()function, the shell variable$1is directly interpolated into the SQL query string (WHERE chunks_fts MATCH '$1'). A malicious search query could be used to execute arbitrary SQL commands within the SQLite database. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and index untrusted external document collections, creating a potential vector for indirect prompt injection.
- Ingestion points: The
build_inventoryandextract_pdf_textfunctions scan and read content from arbitrary files in a user-specified directory. - Boundary markers: None. The extracted text is indexed into SQLite FTS5 without delimiters or safety warnings for the agent.
- Capability inventory: File system reading (
fitz.open), SQLite database creation and writing, and a Bash-based query interface. - Sanitization: No text sanitization, filtering, or instruction-stripping is performed on the content extracted from documents before it is stored or retrieved for the agent's context.
Audit Metadata