Most chat interfaces treat conversations as short-lived; nanobot treats agent work as persistent workflows you can own. It keeps the runtime small and inspectable while providing the pieces needed for long-running goals: chat channels, tools, durable memory, model routing and automation — all with simple self-hosting and extension points.
What Sets It Apart
- Small, inspectable core: the agent loop and runtime are intentionally compact so contributors can understand and modify behavior without a large orchestration layer — so what: faster onboarding for maintainers and safer self-hosting.
- Persistent workflows & memory: session history plus a Dream-backed long-term memory let agents sustain multi-step goals and scheduled automations — so what: you can run long-horizon tasks (cron-style automations, ongoing monitoring, or multi-step planning) without losing context.
- Chat-native reach and provider freedom: bundled WebUI plus connectors for Telegram, Discord, Slack, WeChat, email and more, and OpenAI-compatible provider presets (local LLMs such as vLLM supported) — so what: integrate into existing chat ecosystems or plug in local/hosted models without code changes.
- Developer ergonomics & integration points: Python SDK, OpenAI-compatible API surface, MCP support, and deployment options (PyPI, source, Docker) — so what: embed nanobot in pipelines, build custom skills, or operate it as a local gateway.
Who It's For and Tradeoffs
Great fit if you want a self-hosted personal agent that prioritizes transparency, long-running workflows, and multi-channel chat reach. It is especially useful for developers or teams who need a lightweight agent runtime that can be extended with custom tools, automations, or local models.
Look elsewhere if you need an enterprise-grade, fully managed orchestration platform with heavy-duty scaling guarantees out of the box; nanobot favors a compact, user-auditable core and expects operators to handle production hardening and security configuration themselves.
Where It Fits
Technically sits between toy single-file agents and heavyweight orchestration platforms: a practical runtime for owners who want control over models, tooling and memory while retaining an accessible developer surface and chat-first integrations.
