skills/mukul975/anthropic-cybersecurity-skills/implementing-api-security-posture-management/Gen Agent Trust Hub
implementing-api-security-posture-management
Pass
Audited by Gen Agent Trust Hub on Mar 15, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill processes external data from traffic logs and OpenAPI specifications, creating a surface for indirect prompt injection where malicious content in logs could influence agent behavior.
- Ingestion points: The
discover_apis_from_trafficandcheck_api_security_controlsfunctions inscripts/agent.pyingest data from user-provided files. - Boundary markers: No delimiters or boundary markers are utilized to separate untrusted data from processing logic.
- Capability inventory: The script can read local files and write analysis reports to the filesystem.
- Sanitization: Log data is parsed but not sanitized for malicious instruction patterns.
- [EXTERNAL_DOWNLOADS]: The file
references/api-reference.mdincludes documentation examples that reference external APIs and CLI tools from 42Crunch and Salt Security. These are well-known security platforms and the references are provided for educational purposes. - [SAFE]: The core functionality of the skill involves local analysis of API patterns and risk scoring, which is performed without the use of dynamic execution or unauthorized network communication.
Audit Metadata