finance-markets-research

Pass

Audited by Gen Agent Trust Hub on Aug 14, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses a bash script (scripts/crawlora.sh) to wrap curl calls to the Crawlora API. The script follows defensive programming practices, including set -euo pipefail and proper array handling for arguments, effectively preventing common command injection vulnerabilities.
  • [CREDENTIALS_SAFE]: API authentication is handled via the CRAWLORA_API_KEY environment variable. This is a standard and recommended practice for secret management, avoiding hardcoded credentials in code or documentation.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests external data (JSON) from various financial platforms via the Crawlora API.
  • Ingestion points: API responses in scripts/crawlora.sh from api.crawlora.net.
  • Boundary markers: None explicitly defined in the script; the data is returned as raw JSON.
  • Capability inventory: Subprocess execution via curl in scripts/crawlora.sh.
  • Sanitization: Standard JSON handling via jq is suggested in examples. While external data always poses a theoretical risk of indirect prompt injection, the risk is low given the structured nature of the financial data and the use of a managed vendor API.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 14, 2026, 06:10 PM
Security Audit — agent-trust-hub — finance-markets-research