# DocuSOR > DocuSOR is shared, verified context and memory for coding agents. It keeps what agents know about your codebase outside any one agent, tool, or model vendor, so every agent and every engineer on your team reads the same facts through MCP. Every fact is proven against your running app and re-validated as code changes. Measured on Claude Opus 5.5: same quality, 77% faster, 64% lower model cost, 82% fewer tokens, 95% fewer tool calls. ## Why DocuSOR instead of agent-managed memory - Agent- and model-agnostic: works with Claude Code, Cursor, GitHub Copilot, Codex, Windsurf, Cline, and any MCP-capable agent. Switch or mix tools and nothing is lost. - Shared: one context and memory for every agent and every engineer, not per-tool or per-person files like CLAUDE.md, Copilot memory, or Cursor rules. What one agent learns and proves, every other agent reads. - Faster and cheaper: one verified read replaces dozens of exploration tool calls. Benchmark: 152.7s to 34.6s and $0.53 to $0.19 model cost, at equal audit quality (92.4 vs 92.2). DocuSOR reads cost $0.10 per credit. - Reliable: deterministic checks, no LLM deciding what is true. Context whose checks fail or drift is marked stale instead of being served as current. ## Product - [Home](https://docusor.ai/): Overview, benchmark results, pricing, and a net savings calculator. - [Documentation](https://docusor.ai/docs.md): How it works, getting started, MCP setup, GitHub and GitLab access, and credits. - [MCP setup](https://docusor.ai/docs#connect-your-coding-agent-mcp-setup): Server URL, auth header, and configs for Claude Code, Cursor, VS Code, and Windsurf. - [FAQ](https://docusor.ai/faq.md): What DocuSOR is, how it differs from built-in agent memory, billing, and security. - [Case study](https://docusor.ai/case-studies): Methodology and results of the Claude Opus 5.5 head-to-head audit benchmark. - [Guide: Reduce AI agent token costs](https://docusor.ai/guides/reduce-ai-agent-token-costs.md): Where agent tokens go and how to estimate savings. - [Guide: Writing testable docs](https://docusor.ai/guides/testable-docs.md): How to write README and runbook claims DocuSOR proves against your running app. - [Security](https://docusor.ai/security): Data handling, access scopes, and self-hosting. - [Enterprise](https://docusor.ai/docs.md): Annual platform fee plus prepaid credits; unlimited repositories and seats, GitHub and GitLab (GitLab.com or self-managed), SSO over SAML or OIDC, teams, self-hosting including air-gapped, private runners, audit logs, and advanced support. See the Enterprise section of the docs. - [Changelog](https://docusor.ai/changelog.md): Product updates. - [Status](https://docusor.ai/status): Live service status. ## Legal - [Privacy Policy](https://docusor.ai/privacy) - [Terms of Service](https://docusor.ai/terms)