Stop copy-pasting context between apps. Give your AI direct access to your prompts, documents, and collections with a single configuration.
Every conversation starts from zero. Your best prompts live in one app, your research in another, and your AI can't reach any of it.
Say it in plain language. Your AI calls the right tools behind the scenes.
"Use my PR review prompt on this diff"
Your AI calls get_prompt, retrieves your review template with all its variables, and applies it. No copy-paste.
"What did my research say about rate limiting?"
Your AI calls deep_search, finds the relevant paragraphs across all your documents, and synthesizes an answer with sources.
"Save this conversation as a doc in my Architecture collection"
Your AI calls create_document then add_to_collection. Two tools, one sentence. Done.
"Roll back my API guidelines to last Tuesday's version"
Your AI calls get_document_versions to find the right snapshot, then restore_document_version to bring it back.
One click for Cursor. One config block for Claude. Under 60 seconds either way.
Full CRUD with version history, semantic search, and deep document exploration. Everything your AI needs to manage and search your knowledge base.
list_promptsList all prompts with optional search
get_promptGet full prompt details including content and variables
create_promptCreate a new prompt template with ${variables}
update_promptUpdate an existing prompt with version tracking
delete_promptPermanently delete a prompt
get_prompt_versionsView full version history with change logs
restore_prompt_versionRestore a prompt to any previous version
Most tools only search by title or filename. Deep Search lets your AI search within document content, find the exact paragraph it needs, and navigate through your documents like a table of contents -- progressively exploring without loading everything at once.
No other MCP server offers hierarchical vector search inside documents.
Vector-powered content search
Finds relevant passages by meaning, not just matching keywords. Ask a question and get the specific paragraphs that answer it.
Hierarchical navigation
Move up to parent sections, down into paragraphs, or sideways through siblings. Five directions of navigation from any search result.
Session-aware deduplication
Repeated searches in the same session surface new results every time — your AI never gets the same content twice.

Generate a key, add the config, restart your client
Go to Settings → API Keys in your Context Repo dashboard and create a new key.
Copy the config JSON and add it to your client's MCP configuration file. Replace gm_your_api_key with your actual key.
Fully quit and reopen your MCP client. Look for the tools indicator to confirm the connection is active.
{
"mcpServers": {
"context-repo": {
"command": "npx",
"args": ["-y", "context-repo-mcp"],
"env": {
"CONTEXTREPO_API_KEY": "gm_your_api_key"
}
}
}
}
The free tier includes all 26 MCP tools. Install in under 60 seconds and your AI reads your context directly.