skills/baarai/artex/api-security/Gen Agent Trust Hub

api-security

Pass

Audited by Gen Agent Trust Hub on Nov 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill guides the agent to use several standard security auditing tools, including curl, jq, ffuf, kiterunner, and jwt_tool. These tools are intended for request replay, endpoint discovery, and token analysis within the scope of an authorized API audit.\n- [INDIRECT_PROMPT_INJECTION]: As an API security auditing tool, the skill's workflow involves ingesting and analyzing untrusted data from external API endpoints (e.g., responses, schema definitions, and WebSocket frames). This creates a potential surface for indirect prompt injection if a target API returns malicious content.\n
  • Ingestion points: API responses, GraphQL introspection queries, and WebSocket frame payloads across all ten defined testing phases in SKILL.md.\n
  • Boundary markers: The skill references internal platform mechanisms like "standing finding quality gates" and "adversarial gates" to govern behavior.\n
  • Capability inventory: The skill utilizes network-capable tools (curl, ffuf, jwt_tool) for interacting with external services.\n
  • Sanitization: The skill relies on the agent's internal safety filters and instructs the user/agent to verify findings with manual replays and artifact collection.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 9, 2026, 11:17 PM