ai-tech-fulltext-fetch
Pass
Audited by Gen Agent Trust Hub on Mar 6, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches article content from URLs retrieved from a local SQLite database using Python's
urllib.request.urlopen. It implements timeouts and response size limits to prevent resource exhaustion during network operations. - [COMMAND_EXECUTION]: Interacts with a local SQLite database file to manage article metadata and content. It utilizes the
sqlite3library with parameterized queries to prevent SQL injection and manages database state through initialization and sync commands. - [PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it ingests and stores untrusted content from the web.
- Ingestion points: The
fetch_htmlfunction inscripts/fulltext_fetch.pyretrieves content from arbitrary URLs stored in theentriestable. - Boundary markers: Absent. Extracted text is stored in the
entry_contenttable without specific delimiters or warnings for downstream agents. - Capability inventory: The script can perform network requests and write to the local filesystem via SQLite.
- Sanitization: Uses the
trafilaturalibrary or a customHTMLParserto extract plain text, which removes HTML tags but does not filter the semantic content of the retrieved text.
Audit Metadata