We build Oto — a B2B automation toolkit that gives AI assistants real tools: company & people intelligence, prospecting, CRM, messaging, French open data, and a structured knowledge base. Everything is exposed over the Model Context Protocol (MCP), so it works across Claude, ChatGPT and other LLMs.
The platform is composed, not monolithic — and largely open source, product core included.
| Repo | What |
|---|---|
oto-backend |
The backend — encrypted credential vault, orgs, connectors, doctrine, monitoring. Two faces: MCP + REST. The center of the platform. |
oto-dashboard |
Product dashboard for the backend (Vue 3 + shadcn-vue + Tailwind, Logto PKCE). |
oto |
Architecture doctrine — overview + decisions. Public entry point. |
| Repo | What |
|---|---|
oto-core |
Connector library (oto.tools) |
oto-cli |
Business-automation CLI, for humans and agents |
oto-plugin |
Claude Code plugin — auto-configured MCP connector |
otomata-mcp |
Common base for Otomata MCP servers — doctrines, runs, scoped RBAC, call logging |
france-opendata |
French public-data clients (Sirene, INPI, BODACC, DVF…) |
o-browser |
Python browser automation (Patchright / remote CDP) |
oto-memento |
Memento — knowledge substrate of the platform (extraction from oto-backend in progress) |