{
  "markdown": "# AgentDrive MCP Server\n\nThe official [Model Context Protocol](https://modelcontextprotocol.io) server for\n**[AgentDrive](https://agentdrive.sh)** — a cross-agent artifact workspace with\nprovenance and portable storage across Claude Code, Codex, Cursor, LangGraph, and\nany MCP-aware host.\n\nGive your agents a shared, persistent workspace: files with versioning, semantic\nrecall over everything prior agents produced, run logging, and cross-agent\nattribution — so the next session (on any tool) picks up where the last one left off.\n\n- **Website:** https://agentdrive.sh\n- **Docs:** https://agentdrive.sh/docs\n- **npm:** [`@adrive/mcp-server`](https://www.npmjs.com/package/@adrive/mcp-server)\n- **Hosted endpoint:** `https://mcp.agentdrive.sh/mcp`\n\n## Get an API key (free)\n\nCreate one at **https://app.agentdrive.sh** — no credit card. The free tier\n(1 workspace, 1 GB storage, 10k embeddings) is free indefinitely.\n\n## Two ways to connect\n\n### 1. Hosted (no install) — Streamable HTTP\n\nPoint any MCP client that supports remote servers at the hosted endpoint and pass\nyour key as a bearer token:\n\n```\nURL:    https://mcp.agentdrive.sh/mcp\nHeader: Authorization: Bearer agentdrive_live_xxx\n```\n\nOr install in one click via **[Smithery](https://smithery.ai/servers/rakesh1002/agentdrive)**.\n\n### 2. Local (stdio) — via npx\n\n```jsonc\n{\n  \"mcpServers\": {\n    \"agentdrive\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"@adrive/mcp-server\"],\n      \"env\": {\n        \"AGENTDRIVE_API_KEY\": \"agentdrive_live_xxx\"\n      }\n    }\n  }\n}\n```\n\nAdd that block to your host's MCP config:\n\n- **Claude Desktop:** `claude_desktop_config.json`\n- **Cursor:** `~/.cursor/mcp.json`\n- **Claude Code:** `claude mcp add agentdrive -- npx -y @adrive/mcp-server` (set `AGENTDRIVE_API_KEY` in env)\n- **Codex / Cline / Continue / Zed:** the standard `mcpServers` block above\n\nThe fastest path is the AgentDrive installer, which auto-wires every detected host:\n\n```sh\ncurl -fsSL agentdrive.sh/install | sh\n```\n\n## Configuration\n\n| Env var | Required | Default | Description |\n|---|---|---|---|\n| `AGENTDRIVE_API_KEY` | yes | — | Your AgentDrive API key (`agentdrive_live_...`). |\n| `AGENTDRIVE_API_URL` | no | `https://app.agentdrive.sh` | Override the API endpoint (self-hosted worker). |\n| `AGENTDRIVE_AGENT_PLATFORM` | no | `claude-code` | Which agent runtime is calling — drives cross-agent provenance. One of `claude-code`, `codex`, `cursor`, `langgraph`, `crewai`, `autogen`, `cline`, `continue`, `zed`, `unknown`. |\n| `AGENTDRIVE_AGENT_ROLE` | no | — | Free-text role label (e.g. `planner`, `reviewer`) to distinguish multiple agents on the same platform. |\n\n## Tools\n\n20 tools across files, memory, collaboration, org, and provenance.\n\n| Tool | Category | Description |\n|---|---|---|\n| `drive_context` | memory | Auto-pull the most relevant prior artifacts for the task you're starting. |\n| `drive_search` | memory | Semantic search across your org's artifacts. |\n| `drive_recall` | memory | Recall persistent facts/preferences agents have written. |\n| `drive_write` | files | Write a markdown file (auto-versioned). |\n| `drive_read` | files | Read a file's content. |\n| `drive_list` | files | List files in a workspace. |\n| `drive_delete` | files | Delete a file. |\n| `drive_versions` | files | View a file's version history. |\n| `drive_share` | collaboration | Generate a (optionally email-gated) share link. |\n| `drive_comment` | collaboration | Comment on a file. |\n| `drive_feedback` | collaboration | Read human comments on a file. |\n| `drive_activity` | provenance | Recent workspace activity (writes, comments, shares). |\n| `drive_log_run` | provenance | Log an agent run (task, status, artifacts, identity). |\n| `drive_invite` | org | Invite a human to your org by email. |\n| `drive_roles` | org | Assign/list/remove per-workspace access by email. |\n| `drive_org` | org | View org details: agents, members, plan, usage. |\n| `drive_sandbox_clone` | sandbox (experimental) | Clone a git repo into a per-workspace Linux sandbox. |\n| `drive_sandbox_exec` | sandbox (experimental) | Run a shell command in the sandbox. |\n| `drive_sandbox_read` | sandbox (experimental) | Read a file from the sandbox filesystem. |\n| `drive_sandbox_commit` | sandbox (experimental) | Stage + commit the sandbox worktree. |\n\nThe `drive_sandbox_*` tools require the Cloudflare Containers plan on the AgentDrive\nworker and degrade cleanly to a \"not configured\" message otherwise.\n\n## Why AgentDrive\n\nAgents lose context at every session boundary and every tool switch. AgentDrive is\nthe shared substrate: one workspace, every agent, full provenance. Write once with\nClaude Code; recall it from Cursor; hand off to Codex — each call is attributed to\nthe right principal and platform.\n\n## Develop\n\n```sh\nnpm install\nnpm run build      # tsc -> dist/\nnpm test           # vitest\n```\n\n## License\n\nApache-2.0. See [LICENSE](./LICENSE).\n",
  "bytes": 4894,
  "sha": "a31f6f5557c5cb1c6db8903844c0eb36f8a727f813449b18d266a54c7fe7be7a",
  "repo_slug": "rakesh1002/agentdrive-mcp",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/mcp_sh_agentdrive_agentdrive_3a0e6cb1/readme"
}