The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions involve running shell commands for environment setup and project maintenance.
Commands include pnpm install for dependency management, pnpm run codegen for API client generation, and pnpm run push for Drizzle ORM database migrations (SKILL.md).
[EXTERNAL_DOWNLOADS]: The integration uses standard, well-known libraries from the NPM registry.
Dependencies include @anthropic-ai/sdk, p-limit, and p-retry (templates/lib/integrations-anthropic-ai/package.json).
[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface for indirect prompt injection as it is designed to process and store external AI responses.
Ingestion points: Untrusted content enters the system via the anthropic.messages.stream() call (SKILL.md, Wiring Instructions).
Boundary markers: The provided templates do not include explicit delimiters or safety instructions for the processed content.
Capability inventory: The skill provides functionality to write received content to a PostgreSQL database via Drizzle ORM (lib/db/src/schema/messages.ts) and stream it to the client via SSE.
Sanitization: No specific sanitization or filtering logic is provided in the library templates; implementation is left to the user.

ai-integrations-anthropic