Context+ MCP Server

Content

Context+ is an MCP server for semantic and structural codebase intelligence. It combines AST parsing, semantic embeddings, spectral clustering, Obsidian-style feature hubs, static analysis, restore points, and memory graph tools so Claude can navigate large repositories with less file-by-file context loading.

The upstream README documents npx -y contextplus, bunx contextplus, IDE MCP config generation, Ollama and OpenAI-compatible embedding providers, runtime caches, memory graph tools, and the propose_commit write path with shadow restore points.

Source Review

• https://github.com/forloopcodes/contextplus
• https://github.com/forloopcodes/contextplus/blob/main/README.md
• https://github.com/forloopcodes/contextplus/blob/main/package.json
• https://registry.npmjs.org/contextplus

These sources were reviewed on 2026-06-05. Prefer the live repository, README, package metadata, and NPM registry metadata for current package version, command names, tools, embedding provider settings, cache behavior, and supported agent config targets.

Features

• Return structural AST trees and file skeletons.
• Search code semantically by meaning or identifier.
• Navigate clustered code areas with semantic labels.
• Trace symbol blast radius across files and line ranges.
• Run static analysis for supported language ecosystems.
• Propose code changes with validation and shadow restore points.
• Search and traverse a memory graph with typed relations.
• Generate MCP config files for supported coding agents.
• Cache file, identifier, and call-site embeddings for reuse.

Installation

For MCP clients that launch stdio servers:

{
  "mcpServers": {
    "contextplus": {
      "command": "npx",
      "args": ["-y", "contextplus"],
      "env": {
        "CONTEXTPLUS_EMBED_PROVIDER": "ollama",
        "OLLAMA_EMBED_MODEL": "nomic-embed-text"
      }
    }
  }
}

If using an OpenAI-compatible embedding provider, configure the provider, API key, base URL, embedding model, and optional chat model according to the upstream README. Restart the MCP client after adding the server.

Use Cases

• Ask Claude to find code by concept rather than exact text.
• Inspect API surfaces through file skeletons without reading full bodies.
• Trace every place a symbol is imported or used before refactoring.
• Group a large repository into semantic feature areas.
• Run static analysis before accepting a proposed change.
• Store and retrieve feature notes through memory graph traversal.
• Use shadow restore points to undo a Context+ proposed change.

Safety and Privacy

Context+ can read source files, create persistent caches, invoke static-analysis tools, and write code through propose_commit. Review tool calls carefully in repositories containing proprietary code, customer data, secrets, or production configuration.

Local embedding with Ollama can keep source-derived text on the machine, while OpenAI-compatible providers may receive snippets, identifiers, prompts, and cluster-labeling context. Make the provider choice deliberately, and keep API keys out of committed config files.

Duplicate Check

No forloopcodes/contextplus entry, contextplus package entry, or matching source URL was found in content/mcp.

• Source repositorygithub.com 2026-07-21T05:16:31+00:00
• Documentationgithub.com
• Submitted by oktofeesh12026-06-05