Does code-memory require an API key or internet connection?

No. It runs entirely offline using local embeddings and Git history. You only need a HuggingFace model ID specified in the EMBEDDING_MODEL environment variable.

How do I configure the embedding model?

Set the EMBEDDING_MODEL environment variable to a HuggingFace model ID before running the server. This determines which model generates code embeddings for semantic search.

Can I adjust logging verbosity?

Yes, use the CODE_MEMORY_LOG_LEVEL environment variable and set it to DEBUG, INFO, WARNING, or ERROR depending on how much detail you need.

What Git history does code-memory search?

The server indexes your local Git repository, allowing you to search across commits and historical code changes. Exact depth and filtering depend on your repository configuration.

How do I install and run code-memory?

Install with `uvx code-memory@1.0.28`. It communicates via stdio, so connect it to your MCP client and configure EMBEDDING_MODEL before querying.

MCP ServerSTDIOOfficialv1.0.28

Kapillamba4 Code Memory MCP Server

Search your codebase semantically using Git history and local embeddings—no API keys or internet required. Ideal for developers who need fast, offline code discovery across large repositories.

io.github.kapillamba4/code-memory

Try MCP with AI models View source

Hosted URL

Local install

Transport

STDIO

Auth

No auth required

What the Kapillamba4 Code Memory MCP server does

How models use it and what it is built for.

Search your codebase semantically using Git history and local embeddings—no API keys or internet required. Ideal for developers who need fast, offline code discovery across large repositories.

Connect to Kapillamba4 Code Memory

Local install — runs as a subprocess.

uvx code-memory==1.0.28

Environment variables

Configuration this server reads at startup.

CODE_MEMORY_LOG_LEVEL
Logging verbosity (DEBUG, INFO, WARNING, ERROR)
EMBEDDING_MODEL
HuggingFace model ID for embeddings

Resources

Where to find authoritative docs and source for Kapillamba4 Code Memory.

Source repository

Example prompts for Kapillamba4 Code Memory

Paste any of these into Agent Studio after connecting Kapillamba4 Code Memory.

Find all functions that handle user authentication across the repo
Show me code similar to this database connection pattern
Search for error handling logic in the last 50 commits
What embedding model is currently configured and how do I change it

Kapillamba4 Code Memory MCP server — FAQ

Common questions about connecting and running Kapillamba4 Code Memory.

Does code-memory require an API key or internet connection?
No. It runs entirely offline using local embeddings and Git history. You only need a HuggingFace model ID specified in the EMBEDDING_MODEL environment variable.
How do I configure the embedding model?
Set the EMBEDDING_MODEL environment variable to a HuggingFace model ID before running the server. This determines which model generates code embeddings for semantic search.
Can I adjust logging verbosity?
Yes, use the CODE_MEMORY_LOG_LEVEL environment variable and set it to DEBUG, INFO, WARNING, or ERROR depending on how much detail you need.
What Git history does code-memory search?
The server indexes your local Git repository, allowing you to search across commits and historical code changes. Exact depth and filtering depend on your repository configuration.
How do I install and run code-memory?
Install with `uvx code-memory@1.0.28`. It communicates via stdio, so connect it to your MCP client and configure EMBEDDING_MODEL before querying.

Skip the local setup — run MCP in your browser

MCP Playground runs 10,000+ hosted MCP servers — GitHub, Linear, Notion, Stripe, Sentry and more — across Claude, GPT, Gemini, DeepSeek and 40+ AI models. Compare model answers side-by-side, save agent presets, share runs. Zero install.

Open Agent Studio