Claude MCP Server

Co-Engram adapter for Claude Code (MCP server).

Local serverstdioTypeScript

What is the Claude MCP server?

Co-Engram adapter for Claude Code (MCP server). That is what the claude mcp server brings to an AI assistant: the same capability, reachable through the Model Context Protocol rather than a separate app or dashboard.

The short version

Getting it running

@co-engram/claude-code on npm is all you need. Most clients run it directly, so configuration is a few lines and a restart.

The tools it exposes

The server publishes 14 tools. What each one is for:

  • Always — on second-level injection** — the moment a session starts, a second-level directory overview (each project domain + its sub-domain directories, each
  • Differentiator — What it means
  • Self-maintaining — A maintenance engine runs light (RPE-based reinforcement), deep (consolidation + decay), and rem (metacognition upgrade/refute) stages automatically
  • Tier — Returns
  • catalog — id, title, kind, domainTags
  • digest — catalog + summary, importance, status
  • content — full frontmatter + Markdown body
  • auto — deepest tier that fits contextBudget.totalTokens
  • Trigger — Effect
  • engram_report_failure — Mark a specific retrieval as harmful; increments failedUses
  • Dimension — What it checks
  • Consistency — Does this agree or conflict with other engrams (via contradicts synapses)?

Things to watch

  • It runs with your machine's permissions. That is convenient and also the reason to think about what you point it at before you approve a tool call.
  • With 14 tools registered it takes up a noticeable share of the context window; turn it off in projects that never touch Claude.
  • Keep per-call confirmation enabled while you learn its behaviour; it is the cheapest safeguard you have.

How it compares

This sits in the file and storage access group, where several servers overlap in what they claim to do but differ sharply once you actually set them up. Claude's toolset — Always, Differentiator, Self-maintaining and 11 more — is a fair guide to whether it matches your workflow. It is maintained by yangyang_for_co_engram; worth a glance at recent repository activity before you build anything load-bearing on it.

We check each listing at SyncDev against the project's documentation before it goes live — if something here drifts out of date, it is a bug worth reporting.

Available tools

ToolWhat it does
Alwayson second-level injection** — the moment a session starts, a second-level directory overview (each project domain + its sub-domain directories, each with a cumulative engram count) is injected into the system prompt. Wit
DifferentiatorWhat it means
Self-maintainingA maintenance engine runs light (RPE-based reinforcement), deep (consolidation + decay), and rem (metacognition upgrade/refute) stages automatically — no manual tagging required.
TierReturns
catalogid, title, kind, domainTags
digestcatalog + summary, importance, status
contentfull frontmatter + Markdown body
autodeepest tier that fits contextBudget.totalTokens
TriggerEffect
engram_report_failureMark a specific retrieval as harmful; increments failedUses
DimensionWhat it checks
ConsistencyDoes this agree or conflict with other engrams (via contradicts synapses)?
LongevityHow long has this engram survived without being refuted?
UsageHow often has it been retrieved and confirmed effective?

How to install the Claude MCP server

{
  "mcpServers": {
    "claude-code-1": {
      "command": "npx",
      "args": ["-y", "@co-engram/claude-code"]
    }
  }
}

Add to claude_desktop_config.json, then restart Claude Desktop.

Example prompts to try

  • Use Claude to Always.
  • Use Claude to Differentiator.
  • Use Claude to Self-maintaining.

Frequently asked questions

It connects Claude to MCP-compatible AI assistants such as Claude and Cursor, exposing 14 tools (Always, Differentiator, Self-maintaining, and more) that the assistant can call on your behalf. Instead of copying data back and forth by hand, the assistant works with Claude directly.