parallel-web

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill provides legitimate research functionality via the Parallel Web Systems API (api.parallel.ai), a well-known service for AI-assisted search and extraction.
  • [SAFE]: The skill uses the PARALLEL_API_KEY environment variable for authentication, which is a standard security practice for managing secrets in AI agent skills.
  • [SAFE]: The implementation in scripts/parallel_web.py is straightforward, using official Python libraries (openai, parallel-web) to interact with APIs. It contains no dynamic code execution (eval, exec), suspicious subprocess calls, or privilege escalation attempts.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8). It retrieves content from untrusted external web pages and incorporates this data into the agent's context, which could potentially contain malicious instructions. Evidence Chain: 1. Ingestion points: Web data processed in scripts/parallel_web.py through search, research, and extract commands. 2. Boundary markers: None explicitly enforced in the script output. 3. Capability inventory: The agent has access to Read, Write, Edit, and Bash tools. 4. Sanitization: No content filtering or sanitization of retrieved web data is performed.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 11:31 PM
Security Audit — agent-trust-hub — parallel-web