The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions for installing the official Anthropic SDKs and related utilities via standard package registries (e.g., pip install anthropic, npm install @anthropic-ai/sdk, composer require anthropic-ai/sdk). These references target a well-known service and are integral to the skill's intended purpose.
[INDIRECT_PROMPT_INJECTION]: The documentation identifies and describes architectural patterns for building AI agents that ingest untrusted data from external sources (such as via the web_search, web_fetch, and read tools). While this introduces an indirect prompt injection surface, the skill comprehensively documents mitigation strategies, including implementing restrictive permission policies, validating inputs, and using isolated execution sandboxes.
[SAFE]: Code snippets provided for credential management use standard best practices, such as loading secrets from environment variables or dedicated vaults, and utilize benign placeholders for API keys and tokens. No malicious patterns, obfuscation, or unauthorized data access were detected.

claude-api