Build production LLM apps with 2 dependencies.
Async-native RAG, Agents, and Graph Workflows — no magic, no SaaS, no bloat.
Website · Docs · Quickstart · Discord · Contributing
fromsynapsekitimportRAGrag=RAG(model="gpt-4o-mini", api_key="sk-...")
rag.add("Your document text here")
asyncfortokeninrag.stream("Summarise this."):
print(token, end="", flush=True)Streaming RAG in 4 lines.
Files & Docs
Cloud Storage
Databases
Productivity & CRM
Communication & Social
Web & Research
Media
Version 2.0 focuses on trust and autonomy in production, plus a repo-wide hardening pass (42 audited security, reliability, and performance fixes).
- Verifiable Agents — cryptographically signed, hash-chained audit trails (RFC 6962 Merkle, Ed25519 + KMS/BYOK) with a standalone verifier.
- Living Memory — agents propose signed, diffable patches to their memory files instead of overwriting them.
- Property Graph RAG & WorldModelRAG — graph-fused and temporal/causal retrieval.
- Personal Knowledge Mesh — local-first indexing across every project, with a CLI and MCP tools.
- AgentSwarm, SelfImprovingAgent, NeuroSymbolicAgent, and an EdgeRuntime for local-first inference.
v2.0.1 (additive, no breaking changes) adds SynapseKit Live (a zero-dependency real-time run dashboard), official Docker images, OKF support, the Universal Memory Protocol, Digital Twin and Time-Travel agents, a signed agent marketplace, and a security pass to 0 known vulnerabilities.
Upgrading? See the Migrating to 2.0 guide.
| Repo | What it is |
|---|---|
| SynapseKit/SynapseKit | The core library — RAG, agents, graph workflows, 35 LLM providers |
| SynapseKit/synapsekit-docs | Documentation site (Docusaurus), live at synapsekit.github.io/synapsekit-docs |
| SynapseKit/synapsekit-ui | Marketing website, live at synapse-kit.com |
| SynapseKit/evalci | Hosted eval runner — run SynapseKit eval suites in CI via GitHub Action |
"LangChain for people who hate LangChain."
| SynapseKit | LangChain | LlamaIndex | |
|---|---|---|---|
| Hard deps | 2 | 50+ | 20+ |
| Install size | ~5 MB | ~200 MB+ | ~100 MB+ |
| Async-native | ✅ Default | ||
| Streaming | ✅ Default | ||
| Cost tracking | ✅ Built-in | ❌ SaaS | ❌ No |
| Evaluation / EvalCI | ✅ CLI + GitHub Action | ❌ SaaS | |
| Graph workflows | ✅ Built-in | ❌ No | |
| Agent federation | ✅ Built-in | ❌ No | ❌ No |
| Verifiable audit trails | ✅ Signed, hash-chained | ❌ No | ❌ No |
| Reasoning LLMs | ✅ Unified adapter | ||
| Structured output | ✅ Provider-agnostic | ||
| Agent memory | ✅ 4 built-in backends | ||
| Observability | ✅ Prometheus + Grafana | ❌ No | ❌ No |
| Stack traces | Your code | Framework internals | Framework internals |
| License | Apache 2.0 | MIT | MIT |
pip install synapsekit[openai]fromsynapsekitimportRAGrag=RAG(model="gpt-4o-mini", api_key="sk-...")
rag.add("Your document text here")
print(rag.ask_sync("Summarise this."))Full docs → synapsekit.github.io/synapsekit-docs
Browse good first issues · Read the Contributing Guide · Join Discord
⭐ Star the repo if SynapseKit saves you time.