Infimium

Name: Infimium
Author: aryankumar06

by aryankumar06communityv0.1.6

Infimium is a local MCP server that indexes a codebase and local docs to provide semantic code search, dependency graph lookup, project memory, planning, and context handoff for AI agents. It can also optionally do web search and fetch URLs, while keeping indexing and embeddings local via Ollama and ChromaDB.

From the registry: Private context layer for AI agents: web, docs, code, deps, memory, and plans.

GitHub Repository

Install & configure

Prerequisites

required
Install Node.js and npm
Node.js and npm are required to run the Infimium package via npx and to use the doctor/index commands.
required
Install Docker
Docker is the recommended setup path and is used to run ChromaDB locally.
required
Install Ollama
Ollama must be installed and running locally for embeddings.
required
Pull the Ollama embedding model
Infimium requires the local Ollama embedding model nomic-embed-text.
```
$ ollama pull nomic-embed-text
```
required
Start ChromaDB
A local ChromaDB instance must be available, typically via Docker, before using the server.
```
$ docker compose up -d chromadb
```
required
Start Ollama server
Ollama must be running locally so Infimium can connect to it.
```
$ ollama serve
```
optional
Optionally obtain a Tinyfish search API key
Needed only if you want the web_search tool. Other tools work without it.

Install in your client

Paste into the AI you're chatting with — it will figure out the rest

Please install the `infimium` MCP server into my current AI client (that's you).

Required prerequisites (do these first if not already done):
- **Install Node.js and npm** — Node.js and npm are required to run the Infimium package via npx and to use the doctor/index commands.
- **Install Docker** — Docker is the recommended setup path and is used to run ChromaDB locally.
- **Install Ollama** — Ollama must be installed and running locally for embeddings.
- **Pull the Ollama embedding model** — Infimium requires the local Ollama embedding model nomic-embed-text. Run: `ollama pull nomic-embed-text`
- **Start ChromaDB** — A local ChromaDB instance must be available, typically via Docker, before using the server. Run: `docker compose up -d chromadb`
- **Start Ollama server** — Ollama must be running locally so Infimium can connect to it. Run: `ollama serve`

Optional prerequisites:
- Optionally obtain a Tinyfish search API key — Needed only if you want the web_search tool. Other tools work without it.

Canonical MCP server config (stdio transport):
- command: `npx`
- args: ["infimium","serve"]
- required environment variables:
  - `CODEBASE_PATH`: Absolute path to the repository Infimium should index and serve. (example: `<absolute-path-to-your-repo>`)
  - `LOCAL_DOCS_PATH`: Path to local documentation to index. The README examples point to the repo docs directory. (example: `<absolute-path-to-your-repo-docs>`)
  - `CHROMADB_HOST`: URL of the local ChromaDB instance used for vector search. (example: `<your-chromadb-url>`)
  - `OLLAMA_HOST`: URL of the local Ollama instance used for embeddings. (example: `<your-ollama-url>`)
  - `SHELL_ALLOWLIST`: Comma-separated list of shell commands the shell tool may run. (example: `<comma-separated-allowed-commands>`)
- optional environment variables:
  - `SEARCH_PROVIDER`: Search provider for the web_search tool. README shows tinyfish as the supported example. (example: `<your-search-provider>`)
  - `SEARCH_API_KEY`: API key for the configured search provider. Required only if using web_search. (example: `<your-search-api-key>`)
  - `INFIMIUM_AUTO_INDEX`: Enable automatic indexing while the MCP server runs. (example: `<true-or-false>`)

Note: Preferred portable config uses npx rather than cloning the repo. The README's setup script also creates a .env, starts ChromaDB and Ollama, pulls the embedding model, indexes the repo, and prints MCP config. If the agent runs from a different workspace than the MCP server, pass project_path once so Infimium can remember the active project.

Add this MCP server to my current client's config in the correct format for you. If you need secrets or credentials I haven't provided, ASK me — do not invent values or leave raw placeholders. After adding it, tell me how to verify the server is connected.

Note: Preferred portable config uses npx rather than cloning the repo. The README's setup script also creates a .env, starts ChromaDB and Ollama, pulls the embedding model, indexes the repo, and prints MCP config. If the agent runs from a different workspace than the MCP server, pass project_path once so Infimium can remember the active project.

Tools (10)

hello_infimium

Health check that returns a simple greeting.

web_search

Search the web using the configured Tinyfish provider.

fetch_url

Fetch a URL and extract readable Markdown or text.

query_local_docs

Search indexed local documentation files such as .md, .txt, .html, and .pdf.

semantic_code_search

Search code by meaning using symbols and local embeddings.

dep_graph

Find where a symbol is defined, its importers, and its imports.

shell

Run allowlisted shell commands with timeout and output limits.

plan

Build an implementation plan from code search and dependency graph context.

project_memory

Save or resume task notes across chats and IDEs.

get_context

Return compact repository context including task, memory, index health, git changes, and touched files.

Environment variables

SEARCH_API_KEYrequiredAPI key for the optional Tinyfish web search provider.

Packages

infimium

npmv0.1.6transport: stdio

Details

Language: TypeScript
License: MIT
Released: Jul 18, 2026
Last commit: Jul 18, 2026
Registry name: io.github.aryankumar06/infimium

Related Servers

Claude Flow

AI orchestration with hive-mind swarms, neural networks, and 87 MCP tools for enterprise dev.

68.0k

Codebase Memory Mcp

Persistent codebase knowledge graph. Survives session restarts and context compaction.

39.2k

Serena

A powerful toolkit for coding, providing semantic retrieval and editing capabilities.

28.1k