{"title":"What is MCP?","slug":"what-is-mcp","category":"mcp","summary":"Model Context Protocol lets LLMs call external tools. Synapse exposes 79 tools via official MCP server.","audience":["human","llm"],"tags":["mcp","intro","overview"],"difficulty":"beginner","updated":"2026-06-27","word_count":565,"read_minutes":3,"llm_context":"MCP = Model Context Protocol (Anthropic, 2024). Open standard for LLM-tool integration.\nSynapse has official MCP server: synapse-mcp-api (npm package, npx -y synapse-mcp-api@latest)\n79 tools exposed: 22 memory, 7 chat, 8 scheduler, 4 tasks, 5 scripts, 9 computers, 4 push, 5 user, 3 utility\n3 transports: stdio (local), HTTP/SSE (remote), WebSocket (mobile)\nSupported clients: Claude Desktop, Claude Code, Cursor, Continue, Cline, any MCP-compatible client\nTool Profiles (v1.4.0): minimal (8 tools), standard (25), full (119) — controlled via MCP_PROFILE env or Mcp-Tool-Profile header\n","lang":"en","translated":true,"requested_lang":"en","content_markdown":"\n# What is MCP?\n\nThe **Model Context Protocol (MCP)** is an open standard by Anthropic (2024)\nthat lets LLMs call external tools in a structured way. Instead of pasting API\ndocs into the prompt, you register tools with the MCP server, and the LLM\ncalls them as needed — like function calling, but standardized and\nclient-agnostic.\n\n## Synapse MCP Server\n\nSynapse ships an official MCP server (`synapse-mcp-api` on npm) that exposes\n**79 tools** covering all Synapse features:\n\n| Category | Tools | Count |\n|----------|-------|-------|\n| Memory | recall, list, store, search, semantic-search, update, delete, bulk-delete, stats, unverified, contradictions, audit, related, by-tag, diff, expiring, health, sync, embed-batch, verify, unverify, mind-export | 22 |\n| Chat | poll, reply, status, history, unread, send, upload | 7 |\n| Scheduler | cron_list, cron_create, cron_delete, cron_toggle, var_list, var_get, var_set, var_delete | 8 |\n| Tasks | task_list, task_get, task_create, task_update | 4 |\n| Scripts | script_list, script_get, script_info, script_store, script_delete | 5 |\n| Computers | computer_list, computer_get, install_code, screenshot, command_queue, command_status, commands_list, disable, delete | 9 |\n| Push | vapid_public_key, subscribe, unsubscribe, test | 4 |\n| User/Mind | register, login, minds_list, mind_create, mind_delete | 5 |\n| Utility | time, calc, random | 3 |\n| Visualization | graph, tags, compact | 3 |\n| Sharing | share, list, revoke | 3 |\n| Webhooks | register, list, get, update, delete | 5 |\n| Browser | new, navigate, click, type, screenshot, close | 6 |\n| **Total** | | **79+** |\n\n## How It Works\n\n```\n┌──────────────────┐ MCP protocol ┌──────────────────┐ HTTP ┌──────────┐\n│ LLM Client │ ◀─────────────────▶│ Synapse MCP │ ─────────▶ │ Synapse │\n│ (Claude/Cursor) │ (stdio/SSE/WS) │ Server │ │ API │\n└──────────────────┘ └──────────────────┘ └──────────┘\n```\n\n1. You configure your LLM client (Claude Desktop, Cursor, etc.) to use the Synapse MCP server\n2. The client starts the MCP server (via `npx -y synapse-mcp-api@latest`)\n3. The MCP server connects to Synapse API using your Mind Key\n4. The LLM sees all 79 tools as native functions it can call\n5. When the LLM needs to remember something, it calls `memory_store` — the MCP server translates this to `POST /memory` on Synapse\n\n## Transports\n\nThe Synapse MCP server supports three transports:\n\n### stdio (local, recommended for desktop)\n\n```json\n{\n \"mcpServers\": {\n \"synapse\": {\n \"command\": \"npx\",\n \"args\": [\"-y\", \"synapse-mcp-api@latest\"],\n \"env\": {\n \"SYNAPSE_MIND_KEY\": \"mk_...\",\n \"SYNAPSE_URL\": \"https://synapse.schaefer.zone\"\n }\n }\n }\n}\n```\n\n### HTTP/SSE (remote, multi-tenant)\n\nConnect your MCP client to:\n\n```\nURL: https://synapse-mcp.schaefer.zone/sse\nHeaders: Authorization: Bearer YOUR_MIND_KEY\n```\n\n### WebSocket (mobile, high-volume)\n\n```\nURL: wss://synapse-mcp.schaefer.zone/ws?mind_key=YOUR_MIND_KEY\n```\n\n## Tool Profiles (v1.4.0)\n\nTo reduce token overhead for smaller LLMs, the MCP server supports three\ntool profiles:\n\n| Profile | Tools | Tokens | Best For |\n|---------|-------|--------|----------|\n| `minimal` | 8 (composite dispatch) | ~500 | Self-hosted LLMs with ≤8k context |\n| `standard` | 25 (named) | ~2,500 | Mid-size LLMs (Claude Haiku, GPT-3.5) |\n| `full` | 119 (all) | ~8,250 | Large LLMs (Claude Sonnet/Opus, GPT-4) — default |\n\nControl via:\n\n- Env var: `MCP_PROFILE=minimal|standard|full`\n- Header: `Mcp-Tool-Profile: minimal|standard|full`\n\n## Supported Clients\n\n- [Claude Desktop](/docs/mcp/claude-desktop) — Anthropic's desktop app\n- [Claude Code](/docs/mcp/claude-code) — terminal coding agent\n- [Cursor](/docs/mcp/cursor) — AI-powered IDE\n- [Continue.dev](/docs/mcp/continue) — open-source AI coding assistant\n- [Cline](/docs/mcp/claude-code) — VS Code extension\n- Any MCP-compatible client\n\n## Why Use MCP Instead of Direct API?\n\n| Approach | Pros | Cons |\n|----------|------|------|\n| Direct API | Simple, no extra layer | LLM needs to know URLs, headers, auth |\n| MCP | LLM sees native tools, no URL memorization | Extra MCP server process |\n\nFor most LLM agent use cases, MCP is the better choice — the LLM doesn't need\nto remember API paths or auth patterns.\n\n## Next Steps\n\n- [Claude Desktop Setup](/docs/mcp/claude-desktop) — 2-minute config\n- [Claude Code Setup](/docs/mcp/claude-code) — terminal integration\n- [Custom MCP Client](/docs/mcp/custom-client) — build your own\n","content_html":"

What is MCP?

\n

The Model Context Protocol (MCP) is an open standard by Anthropic (2024)\nthat lets LLMs call external tools in a structured way. Instead of pasting API\ndocs into the prompt, you register tools with the MCP server, and the LLM\ncalls them as needed — like function calling, but standardized and\nclient-agnostic.

\n

Synapse MCP Server

\n

Synapse ships an official MCP server (synapse-mcp-api on npm) that exposes\n79 tools covering all Synapse features:

\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
CategoryToolsCount
Memoryrecall, list, store, search, semantic-search, update, delete, bulk-delete, stats, unverified, contradictions, audit, related, by-tag, diff, expiring, health, sync, embed-batch, verify, unverify, mind-export22
Chatpoll, reply, status, history, unread, send, upload7
Schedulercron_list, cron_create, cron_delete, cron_toggle, var_list, var_get, var_set, var_delete8
Taskstask_list, task_get, task_create, task_update4
Scriptsscript_list, script_get, script_info, script_store, script_delete5
Computerscomputer_list, computer_get, install_code, screenshot, command_queue, command_status, commands_list, disable, delete9
Pushvapid_public_key, subscribe, unsubscribe, test4
User/Mindregister, login, minds_list, mind_create, mind_delete5
Utilitytime, calc, random3
Visualizationgraph, tags, compact3
Sharingshare, list, revoke3
Webhooksregister, list, get, update, delete5
Browsernew, navigate, click, type, screenshot, close6
Total79+
\n

How It Works

\n
┌──────────────────┐    MCP protocol    ┌──────────────────┐    HTTP    ┌──────────┐\n│  LLM Client      │ ◀─────────────────▶│  Synapse MCP     │ ─────────▶ │ Synapse  │\n│ (Claude/Cursor)  │   (stdio/SSE/WS)   │  Server          │            │ API      │\n└──────────────────┘                    └──────────────────┘            └──────────┘
    \n
  1. You configure your LLM client (Claude Desktop, Cursor, etc.) to use the Synapse MCP server
    2. \n
  2. The client starts the MCP server (via npx -y synapse-mcp-api@latest)
    3. \n
  3. The MCP server connects to Synapse API using your Mind Key
    4. \n
  4. The LLM sees all 79 tools as native functions it can call
    5. \n
  5. When the LLM needs to remember something, it calls memory_store — the MCP server translates this to POST /memory on Synapse
    6. \n
\n

Transports

\n

The Synapse MCP server supports three transports:

\n

stdio (local, recommended for desktop)

\n
{\n  "mcpServers": {\n    "synapse": {\n      "command": "npx",\n      "args": ["-y", "synapse-mcp-api@latest"],\n      "env": {\n        "SYNAPSE_MIND_KEY": "mk_...",\n        "SYNAPSE_URL": "https://synapse.schaefer.zone"\n      }\n    }\n  }\n}

HTTP/SSE (remote, multi-tenant)

\n

Connect your MCP client to:

\n
URL: https://synapse-mcp.schaefer.zone/sse\nHeaders: Authorization: Bearer YOUR_MIND_KEY

WebSocket (mobile, high-volume)

\n
URL: wss://synapse-mcp.schaefer.zone/ws?mind_key=YOUR_MIND_KEY

Tool Profiles (v1.4.0)

\n

To reduce token overhead for smaller LLMs, the MCP server supports three\ntool profiles:

\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
ProfileToolsTokensBest For
minimal8 (composite dispatch)~500Self-hosted LLMs with ≤8k context
standard25 (named)~2,500Mid-size LLMs (Claude Haiku, GPT-3.5)
full119 (all)~8,250Large LLMs (Claude Sonnet/Opus, GPT-4) — default
\n

Control via:

\n\n

Supported Clients

\n\n

Why Use MCP Instead of Direct API?

\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
ApproachProsCons
Direct APISimple, no extra layerLLM needs to know URLs, headers, auth
MCPLLM sees native tools, no URL memorizationExtra MCP server process
\n

For most LLM agent use cases, MCP is the better choice — the LLM doesn't need\nto remember API paths or auth patterns.

\n

Next Steps

\n\n","urls":{"html":"/docs/mcp/what-is-mcp","text":"/docs/mcp/what-is-mcp?format=text","json":"/docs/mcp/what-is-mcp?format=json","llm":"/docs/mcp/what-is-mcp?format=llm"},"translations_available":["en","zh","hi","es","fr","ar","pt","ru","ja","de","it","ko","nl","pl","tr","sv","vi","th","id","uk"]}