substrate-memory

substrate-memory

Provides Claude Co-work with MCP tools to search and record organizational memory, including search, read, query, ingest, remember, sync, and status operations.

Category
Visit Server

README

Claude Co-work Substrate Memory

Beta — stdio transport unverified in the Co-work sandbox. The plugin format is verified against Anthropic's cached Co-work authoring skill, but its stdio MCP execution environment in Co-work's VM sandbox is not verified. A human must install the .plugin, open a Co-work session, and confirm substrate_status returns before support is claimed.

Substrate Memory gives Claude Co-work seven MCP tools for searching and recording organizational memory. Capture is skill-driven: the bundled skill directs Claude to search prior work before answering, remember durable user facts and decisions, and sync at natural session boundaries. Hooks are a best-effort bonus only and may never fire in Co-work.

Install

  1. Package the repository as a .plugin file as described in docs/packaging.md.
  2. Install the .plugin file in Claude Co-work.
  3. Set the environment variables below in the process that launches Co-work.
  4. Start a session and run substrate_status.

Configuration

Set these secrets in the process environment only:

SUBSTRATE_API_URL=https://api.substrate.example
SUBSTRATE_API_KEY=...

SUBSTRATE_API_KEY is never written to a config file. Host config files are world-readable plaintext. SUBSTRATE_PYTHON optionally overrides the python executable used for the stdio server.

Sandbox consequences and mitigations:

  1. A loopback http://127.0.0.1 Substrate URL may not resolve from the Co-work sandbox; use an HTTPS URL.
  2. python may not be on the sandbox PATH; set SUBSTRATE_PYTHON to an available executable.
  3. .mcp.json also supports an {"type":"http","url":...} transport as an alternative when a hosted MCP endpoint is available.

Tools

  • substrate_search — search cited organizational memory cards.
  • substrate_read — read a wiki page by path.
  • substrate_query — ask for a cited synthesis.
  • substrate_ingest — submit text for ingestion.
  • substrate_remember — record a durable fact, decision, or preference.
  • substrate_sync — flush the durable local spool.
  • substrate_status — report content-free configuration and delivery status.

Privacy boundary

The plugin reads API credentials from the process environment only, redacts captured text before persistence or transfer, and emits content-free operational diagnostics. It does not read Co-work transcripts; hook events contain only a session boundary signal. Report vulnerabilities through GitHub private security advisories.

See BOUNDARY.md and SECURITY.md for the public commitment and security policy.

Repository map

  • skills/ — the primary Co-work capture instructions and tool reference.
  • src/claude_cowork_memory/ — Co-work runtime, MCP server, hook, and CLI.
  • src/substrate_capture/ — frozen shared client, redaction, spool, event, and MCP core.
  • hooks/ — best-effort Stop and SessionEnd integration.
  • tests/ — core and plugin contract tests.
  • docs/packaging.md — .plugin creation instructions.

MIT © 2026 Sightline Technologies Inc.

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured