Continuum

Name: Continuum
Author: pouyahasanamreji

by pouyahasanamrejicommunityv0.3.0

An MCP server that provides shared project memory and multi-agent orchestration for coding assistants. It stores lessons and dispatch records in a local SQLite database with vector search, and helps clients manage isolated agent worktrees and reserved paths.

From the registry: Shared memory + orchestration for your coding agents. Local-first MCP, vector RAG.

GitHub Repository Website

Install & configure

Prerequisites

required
Run the Continuum server
Start the server locally. The README offers Docker Compose as the easiest path, or bare-metal with Node and pnpm. The MCP endpoint is exposed over Streamable HTTP.
```
$ docker compose -f docker-compose.dev.yml up
```
optional
Optional bare-metal runtime
If you do not use Docker Compose, install Node.js >= 22.12.0 and pnpm >= 10.33+, then start the app locally.
```
$ pnpm install && pnpm dev
```
optional
Optional external embedder
Continuum works out of the box with its default in-process embedder. If you want to override embedding, run an HTTP-based embedder such as Ollama, TEI, or another OpenAI-compatible embeddings service, then set EMBEDDER_URL and related embedder settings.
```
$ ollama pull <embedding-model>
```
optional
Optional Anthropic API key
Needed only for token-count endpoints used by the server-side tokenizer features.

Install in your client

Paste into the AI you're chatting with — it will figure out the rest

Please install the `continuum` MCP server into my current AI client (that's you).

Required prerequisites (do these first if not already done):
- **Run the Continuum server** — Start the server locally. The README offers Docker Compose as the easiest path, or bare-metal with Node and pnpm. The MCP endpoint is exposed over Streamable HTTP. Run: `docker compose -f docker-compose.dev.yml up`

Optional prerequisites:
- Optional bare-metal runtime — If you do not use Docker Compose, install Node.js >= 22.12.0 and pnpm >= 10.33+, then start the app locally. Run: `pnpm install && pnpm dev`
- Optional external embedder — Continuum works out of the box with its default in-process embedder. If you want to override embedding, run an HTTP-based embedder such as Ollama, TEI, or another OpenAI-compatible embeddings service, then set EMBEDDER_URL and related embedder settings. Run: `ollama pull <embedding-model>`
- Optional Anthropic API key — Needed only for token-count endpoints used by the server-side tokenizer features.

Canonical MCP server config (HTTP transport):
- url: `http://127.0.0.1:7776/mcp`

Note: Continuum exposes MCP over Streamable HTTP. If started via Docker Compose dev, the API is shown at http://localhost:6685, so the MCP endpoint would be that base URL plus /mcp. Environment variables supported by the server: PORT (optional, HTTP port), CORS_ORIGIN (optional, allowed origins), ORCHESTRATOR_DB_PATH (optional, SQLite file path), PANEL_REST_ENABLED (optional, enable REST surface for panel), NODE_ENV (optional), ORCHESTRATOR_ALLOW_DESTRUCTIVE_MIGRATE (optional), ANTHROPIC_API_KEY (optional, token-count endpoints), ANTHROPIC_TOKENIZER_MODEL (optional, tokenizer model), EMBEDDER_URL (optional, HTTP embedding endpoint; unset uses built-in in-process embedder), EMBEDDER_MODEL (optional, embedder model id), EMBEDDER_DIM (optional, embedding dimension). No MCP-specific headers or auth are described.

Add this MCP server to my current client's config in the correct format for you. If you need secrets or credentials I haven't provided, ASK me — do not invent values or leave raw placeholders. After adding it, tell me how to verify the server is connected.

Note: Continuum exposes MCP over Streamable HTTP. If started via Docker Compose dev, the API is shown at http://localhost:6685, so the MCP endpoint would be that base URL plus /mcp. Environment variables supported by the server: PORT (optional, HTTP port), CORS_ORIGIN (optional, allowed origins), ORCHESTRATOR_DB_PATH (optional, SQLite file path), PANEL_REST_ENABLED (optional, enable REST surface for panel), NODE_ENV (optional), ORCHESTRATOR_ALLOW_DESTRUCTIVE_MIGRATE (optional), ANTHROPIC_API_KEY (optional, token-count endpoints), ANTHROPIC_TOKENIZER_MODEL (optional, tokenizer model), EMBEDDER_URL (optional, HTTP embedding endpoint; unset uses built-in in-process embedder), EMBEDDER_MODEL (optional, embedder model id), EMBEDDER_DIM (optional, embedding dimension). No MCP-specific headers or auth are described.