skills/popodidi/harvis/source-tenders/Gen Agent Trust Hub

source-tenders

Pass

Audited by Gen Agent Trust Hub on Mar 12, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The scraper (scraper.py) fetches tender listings from the official Taiwan Government E-Procurement website (web.pcc.gov.tw). The implementation explicitly disables SSL certificate verification (verify_ssl=False), which is a security best practice violation that could expose the data fetching process to interception.
  • [PROMPT_INJECTION]: The skill processes untrusted content from external web pages, creating an indirect prompt injection surface.
  • Ingestion points: scripts/scraper.py ingests tender titles, agency names, and metadata from the web.pcc.gov.tw domain.
  • Boundary markers: No specific delimiters or safety instructions are used to separate scraped data from the agent's control context in the produced JSON files.
  • Capability inventory: The skill allows local script execution via uv run and file access via the open command.
  • Sanitization: While scripts/analyze.py correctly escapes HTML entities to prevent cross-site scripting (XSS) in the generated report, it does not sanitize the input for natural language instructions that could influence the agent during subsequent processing steps.
  • [COMMAND_EXECUTION]: The skill defines restricted tool usage for executing its own Python scripts and opening the generated HTML analysis report, which is consistent with its stated purpose.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 12, 2026, 11:38 PM