anthropic-agent-sdk
Pass
Audited by Gen Agent Trust Hub on Jul 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
- [SAFE]: The skill provides informational markdown documentation and code examples for developers using the Anthropic Agent SDK. All code snippets are provided for educational purposes within documentation.
- [EXTERNAL_DOWNLOADS]: The documentation references official installation commands for the Anthropic Agent SDK (e.g.,
npm install @anthropic-ai/claude-agent-sdkandpip install claude-agent-sdk). These point to official and reputable package registries. - [PROMPT_INJECTION]: The directive in
SKILL.md(e.g., "CRITICAL: grep references/") is a technical instruction for the agent to utilize the provided documentation files effectively to avoid hallucinating API details. It does not attempt to bypass safety guardrails or override core instructions. - [DATA_EXFILTRATION]: While documentation files like
secure-deployment.mdmention sensitive file paths (e.g.,.env,~/.ssh), they do so in the context of security best practices, advising users on how to protect these resources from autonomous agents.
Audit Metadata