# Oneda

> Oneda is organisational (organizational) memory for AI agents. It captures decisions, rationale, constraints, outcomes and learnings, and makes them available across AI agents, people and sessions — so agents understand why things are the way they are, not just what exists.

Oneda works with Claude Code, Claude.ai / Claude Desktop (via MCP custom connector) and custom agents via MCP. It complements RAG and enterprise search rather than replacing them: retrieval finds information; Oneda preserves the reasoning around it, with provenance and history.

## Pages

- [Home](https://oneda.app/): Organisational memory for AI agents — overview, demo, governance, Claude Code setup
- [Organisational memory](https://oneda.app/organisational-memory): What organisational memory for AI is, and how it differs from context windows, agent memory and RAG
- [AI agent memory](https://oneda.app/ai-agent-memory): Persistent memory shared across specialised agents, scoped by permissions
- [Claude Code memory](https://oneda.app/claude-code-memory): Persistent project context for Claude Code across sessions
- [MCP memory](https://oneda.app/mcp-memory): Oneda's MCP server for MCP-compatible clients
- [Enterprise](https://oneda.app/enterprise): Security and governance; contact
- [Pricing](https://oneda.app/pricing): Plans
- [Docs](https://oneda.app/docs): Quickstart, connectors, MCP tools reference, imports, API keys, security, troubleshooting

## Getting started

- Sign up: https://oneda.freeelatech.com/oneda.html?signup=1
- Claude Code CLI:
  - `npm install -g @nova-oneda/cli`
  - `nova-oneda setup --key <API_KEY>` (keys begin with `nv_live_`)
- MCP server URL: `https://oneda.freeelatech.com/mcp`
  - Auth header: `authorization: Bearer <API_KEY>`
  - ChatGPT and GitHub Copilot are not supported yet (they require OAuth; Oneda supports Bearer API keys only).

## MCP tools

- `nova_context`: automatically retrieves relevant memory before planning or implementation
- `nova_search`: explicit search across stored memory
- `nova_submit_learning`: records a decision, rationale or insight
- `nova_visualize`: renders the decision graph behind a memory or topic as ASCII
- `nova_import_repo`: imports a GitHub repo's README, manifest, key docs and structure
- `nova_import_jira`: imports recent issues from a Jira project
- `nova_import_confluence`: imports pages from a Confluence space

## Governance

Workspace isolation, API-key access, key revocation, role-based access, memory provenance, memory history, audit logs, encryption at rest.

## Optional

- [Sign in](https://oneda.freeelatech.com/oneda.html)
- [Team / Enterprise sign in](https://oneda.freeelatech.com/index.html)