adapt-custom-endpoint

Pass

Audited by Gen Agent Trust Hub on Sep 1, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to fetch and process external documentation from URLs provided by users to configure API endpoints.
  • Ingestion points: Step 1 in SKILL.md requires the agent to read supplier documentation URLs or user-provided examples.
  • Boundary markers: The skill does not provide delimiters or specific instructions to ignore potential prompt injection attacks that might be embedded within the external documentation content.
  • Capability inventory: The agent has the ability to use the provided Python script to perform network requests (_request in scripts/custom_endpoint.py), execute trial runs which interact with real external providers, and save persistent configurations to the ArcReel backend.
  • Sanitization: No sanitization or verification process is defined for the external documentation content before the agent interprets it for code or JSON generation.
  • [COMMAND_EXECUTION]: The skill executes a local utility script, scripts/custom_endpoint.py, which handles communication with the ArcReel API using standard Python libraries. The script reads local configuration files to authenticate its requests.
  • [EXTERNAL_DOWNLOADS]: The skill performs HTTP requests to external documentation sites and a user-defined API endpoint. The included Python script enforces HTTPS for all non-local connections to ensure the security of API tokens during transmission.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 1, 2026, 12:45 PM
Security Audit — agent-trust-hub — adapt-custom-endpoint