ContextStream MCP Server

Persistent memory. Millisecond code search.

## Your agent knows code. Give it the decisions behind yours. Connect your code, docs, and conversations so your agents can find the right files, recall saved decisions, and build on past work across sessions and tools. ## Start free with MCP **10,000 monthly credits. No credit card required.** Works with Claude Code, Cursor, Codex, GitHub Copilot, Gemini CLI, Qwen Code, Kimi Code, Muse Code, ZCode, Zed, and [more MCP clients](docs/clients.md), on Claude, GPT, Gemini, Kimi, GLM, Qwen, and other models. Create your account or sign in during MCP onboarding. No separate website signup needed. **macOS and Linux** ```sh curl -fsSL https://contextstream.io/scripts/mcp.sh | bash ``` **Windows PowerShell** ```powershell irm https://contextstream.io/scripts/mcp.ps1 | iex ``` Paste the command into your terminal and follow onboarding to connect your project and supported editor. Restart your editor after setup. [What gets installed and indexed?](docs/data-handling.md) · [Supported clients](docs/clients.md) · [Setup guide](https://contextstream.io/docs/editors/setup) · [Prefer to start on the web?](https://contextstream.io/lp/mcp-marketplace?utm_source=github&utm_medium=directory&utm_campaign=mcp_organic_2026q3&utm_content=card_cta) ## Make the next session useful Ask your connected agent: > Use ContextStream to find the files relevant to my next change. Cite the > sources and retrieve any saved project decisions that should guide the work. Then ask it to save a real decision and its reason. Start a new session and retrieve that decision without explaining it again. Indexing supplies code context; it cannot recover every undocumented decision. ## Proof and data controls **95 ms median code search in our published benchmark.** This is the measured successful-response median for the disclosed configuration, not a latency guarantee. [See results, methodology, and limitations](https://contextstream.io/benchmarks). **Scoped access. Traceable sources. Configurable capture.** Indexing sends eligible source contents to ContextStream for hosted search. Transcript saving and local Git metadata capture are on by default and can be turned off. Review [data handling and controls](docs/data-handling.md) before connecting sensitive projects.

Other installation options and troubleshooting

### npm (Node.js 20+) ```bash npx -y @contextstream/mcp-server@latest setup ``` The npm launcher downloads the native Rust binary for your platform and verifies its SHA-256 checksum. Pin an exact package version for production automation. ### Hosted MCP without a local process For clients supporting Streamable HTTP and OAuth, the hosted endpoint is: ```text https://mcp.contextstream.io/mcp?default_context_mode=fast ``` [Supported clients](docs/clients.md) lists one-line add commands for Claude Code, Codex, Gemini CLI, Qwen Code, Copilot CLI, and Droid, and one-click install for VS Code. Use the [MCP documentation](https://contextstream.io/docs/mcp) for other clients and stdio alternatives. A hosted connection alone does not sync your local checkout. ### Preview or diagnose setup ```bash contextstream-mcp setup --dry-run contextstream-mcp doctor --scope=all --only-configured ``` ### Scripted and CI setup For automation, wrap the install in `pipefail` so it fails when the download fails, instead of reporting success after installing nothing: ```bash bash -o pipefail -c 'curl -fsSL https://contextstream.io/scripts/mcp.sh | bash' ``` `setup --yes` reads the key from `CONTEXTSTREAM_API_KEY` or saved credentials, and needs `--workspace-id` when the account has more than one workspace: ```bash export CONTEXTSTREAM_API_KEY=... # or: printf %s "$KEY" | contextstream-mcp configure --api-key-stdin contextstream-mcp setup --yes --editors=claude --workspace-id= --project-path=. ```

## Open source and contributing This repository contains the MIT-licensed Rust MCP server, client, editor setup, and release tooling. The hosted backend is separate and is not included here. See the [architecture](docs/architecture.md), [release integrity](docs/release.md), and [latest release](https://github.com/contextstream/mcp-server/releases/latest). ```bash cargo build --locked -p mcp-server --bin contextstream-mcp cargo test --locked --workspace ``` Use the pinned Rust toolchain. For contribution checks and commit sign-off, read [CONTRIBUTING.md](CONTRIBUTING.md). Report vulnerabilities using [SECURITY.md](SECURITY.md). See [LICENSE](LICENSE), [NOTICE](NOTICE), and [GOVERNANCE.md](GOVERNANCE.md) for licensing, trademarks, and project governance. --- **Let humans be human.** [Docs](https://contextstream.io/docs/mcp) · [Pricing](https://contextstream.io/pricing) · [Integrations](https://contextstream.io/integrations) · [Benchmarks](https://contextstream.io/benchmarks) ### Optional deep project learning Deep project learning starts off for new accounts. Plain `contextstream-mcp setup --yes` preserves your current choice. To review consent, run `contextstream-mcp configure --account-learning on` and confirm on the signed-in privacy page. To withdraw, run `contextstream-mcp configure --account-learning off`. Agents cannot consent for you. Doctor and init show status only. See [data handling and controls](docs/data-handling.md).