aiconfig-tools
Pass
Audited by Gen Agent Trust Hub on May 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill serves as a guide for developers to integrate LaunchDarkly AI Configs into their agents. All external references point to official LaunchDarkly documentation (docs.launchdarkly.com).
- [SAFE]: No sensitive data exposure or exfiltration patterns were identified. The instructions guide the user on how to manage tool schemas and configurations through established Model Context Protocol (MCP) tools.
- [SAFE]: The provided Python code snippets are benign implementation examples for mapping LaunchDarkly tool schemas to various provider SDKs (OpenAI, Anthropic, Bedrock, Gemini).
- [SAFE]: The skill promotes security best practices, such as bounding agent execution loops (MAX_STEPS) and using narrow PATCH updates to prevent accidental data overwrites.
Audit Metadata