Redis AgentKit

A toolkit for building production AI agents entirely on Upstash Redis. It gives you the stateful primitives agents need — memory, conversation history, search, and caching — plus thin adapters for the popular agent frameworks. The "semantic" features are powered by Upstash Redis Search and its $smart fuzzy operator, so no separate vector database is required.

Packages

Core features

• Chat history — durable, Redis-Search-backed conversation transcripts (ChatHistory): save the whole message array per chat, list a user's chats for a sidebar, and fuzzily $smart-search what the user or model said.
• Agent memory — long-term, fuzzily-recalled memories scoped per agent/user, plus drop-in recall/save tools for generateText.
• Search tools — schema-driven search/aggregate/count tools over Upstash Redis Search; the index is created reactively on first use. Use these over your own documents for RAG-style retrieval.
• Rate limiting — a configured Upstash Ratelimit factory (createRateLimit) you call before the model.
• Eve memory slots (Eve only) — Upstash Redis behind Eve's native memory feature: redisDocuments() stores Eve's own fileMemory() documents (so they work off Vercel), and redisMemory() is a full provider with ranked recall and automatic capture.
• Code sandbox (Eve only) — an Upstash Box sandbox provider for Eve's defineSandbox (UpstashSandbox, eve ≥ 0.65).
• Tool-call cache — memoize deterministic tool results keyed by arguments.
• Durable MCP tasks and events (@upstash/mcp-toolkit) — a long-running MCP tool returns a task id instead of blocking, and the model polls task_status; the record lives in Redis and the work runs through QStash or Workflow, so it survives the process that accepted the call. MCP Events let a host subscribe and get a signed webhook when something happens, including when a task finishes.

Examples

Runnable demos (real Upstash Redis + a mock/real model) live in examples/: ai-sdk-demo, eve-demo, eve-extension-demo (an eve agent that mounts @upstash/agentkit-eve-extension), tanstack-ai-demo (a TanStack AI chat whose persistence, resumable stream and memory work across server instances, with a two-instance E2E suite), and mcp-toolkit-demo (an MCP server whose long-running tool returns a task id the model polls, plus a separate events server that fires deploy.finished, with the client's wire log on screen).

Development

pnpm install
pnpm build      # build all packages
pnpm test       # run all tests (against a real Upstash Redis; LLM calls mocked unless OPENAI_API_KEY is set)
pnpm lint       # eslint + prettier
pnpm typecheck  # tsc across packages

Tests need UPSTASH_REDIS_REST_URL / UPSTASH_REDIS_REST_TOKEN (in a repo-root .env); suites that hit Redis skip themselves when absent. Some tests use UPSTASH_BOX_API_KEY and OPENAI_API_KEY.

Telemetry

Each package reports its name and version to Upstash as a header on the requests made by your redis client (e.g. @upstash/redis@1.38.0,@upstash/agentkit-sdk@0.2.0,@upstash/agentkit-ai-sdk@0.2.0), so we know which SDK versions are in use. No personal data, keys or identifiers are collected. Opt out with enableTelemetry: false on any feature config, with the same option on the redis client, or by setting the UPSTASH_DISABLE_TELEMETRY environment variable.

Releasing

This repo uses Changesets.

pnpm changeset        # describe a change
pnpm ci:version       # bump versions + changelogs (also stamps each package's telemetry VERSION constant)
pnpm ci:publish       # publish to npm

# (`version`/`release` script names are avoided — they collide with pnpm's built-in commands.)

License

MIT