leeroopedia-mcp 0.1.2

MCP server for Leeroopedia ML/AI knowledge search

Leeroopedia MCP Server

Give your AI coding agent access to curated ML/AI knowledge.

---

What is Leeroopedia?

Your ML & Data Knowledge Wiki. Learnt by AI, built by AI, for AI. A centralized playbook of best practices and expert-level knowledge for Machine Learning and Data domains.

Browse the full knowledge base at leeroopedia.com. Apply for early beta access.

This MCP server lets AI coding agents (Claude Code, Cursor) search that knowledge base directly while they work — no copy-pasting needed.

---

Quick Start

1. Install

pip install leeroopedia-mcp

2. Get Your API Key

1. Go to app.leeroopedia.com
2. Create an account or log in
3. Navigate to Dashboard > API Keys
4. Copy your API key (format: kpsk_...)

3. Configure Claude Code

Add to your ~/.claude.json or project .mcp.json:

{
  "mcpServers": {
    "leeroopedia": {
      "command": "leeroopedia-mcp",
      "env": {
        "LEEROOPEDIA_API_KEY": "kpsk_your_key_here"
      }
    }
  }
}

4. Configure Cursor

Add to your Cursor settings (.cursor/mcp.json):

{
  "mcpServers": {
    "leeroopedia": {
      "command": "leeroopedia-mcp",
      "env": {
        "LEEROOPEDIA_API_KEY": "kpsk_your_key_here"
      }
    }
  }
}

---

Available Tools

The MCP server provides 8 agentic tools. Each tool (except get_page) triggers an AI agent on the backend that searches the knowledge base from multiple angles, reads relevant pages, and synthesizes a structured response.

Tool	What it does
consult_literature	Research a topic with synthesized, cited answers
build_plan	Build a step-by-step ML execution plan
review_plan	Review a plan against KB best practices
verify_code_math	Verify code against authoritative math/ML descriptions
diagnose_failure	Diagnose training/deployment failures
propose_hypothesis	Propose ranked next-step hypotheses
query_hyperparameter_priors	Query hyperparameter values, ranges & heuristics
get_page	Retrieve a specific KB page by ID

consult_literature — Research a topic like a librarian

Search the knowledge base like a research librarian. An AI agent synthesizes a consensus answer grounded in evidence, with [PageID] citations.

Parameter	Required	Description
query	Yes	Research question or topic to investigate
context	No	Additional context to guide the search

build_plan — Build a step-by-step ML execution plan

Build a step-by-step ML execution plan grounded in knowledge base evidence. Returns an overview, key specs, numbered steps, and validation criteria.

Parameter	Required	Description
goal	Yes	What you want to accomplish
constraints	No	Constraints or requirements (e.g., hardware limits, time budget)

review_plan — Review a plan against best practices

Review a proposed ML plan against knowledge base best practices. Returns approvals, risks, and improvement suggestions.

Parameter	Required	Description
proposal	Yes	The plan or proposal to review
goal	Yes	The intended goal of the plan

verify_code_math — Verify code against ML/math concepts

Verify code correctness against authoritative ML/math concept descriptions. Returns a Pass/Fail verdict with analysis.

Parameter	Required	Description
code_snippet	Yes	The code to verify
concept_name	Yes	The mathematical/ML concept being implemented

diagnose_failure — Diagnose training/deployment failures

Diagnose ML training or deployment failures using knowledge base evidence. Returns diagnosis, fix steps, and prevention advice.

Parameter	Required	Description
symptoms	Yes	Description of the failure symptoms
logs	Yes	Relevant log output or error messages

propose_hypothesis — Propose ranked research hypotheses

Propose ranked research hypotheses grounded in knowledge base evidence. Returns ranked ideas with rationale and suggested experiments.

Parameter	Required	Description
current_status	Yes	Where the project stands now
recent_experiments	No	Description of recent experiments and their outcomes

query_hyperparameter_priors — Query hyperparameter heuristics

Query documented hyperparameter values, ranges, and tuning heuristics. Returns a suggestion table with KB-grounded justification.

Parameter	Required	Description
query	Yes	Hyperparameter question (e.g., "learning rate for LoRA fine-tuning Llama-3 8B")

get_page — Retrieve a KB page by ID

Retrieve the full content of a specific knowledge base page by its exact ID. A direct lookup — no AI agent needed.

Parameter	Required	Description
page_id	Yes	Exact page ID (e.g., Workflow/QLoRA_Finetuning, Principle/LoRA_Rank_Selection)

---

How It Works

The MCP server uses an async task-based API:

1. Your agent calls a tool (e.g., consult_literature)
2. The MCP client sends POST /v1/search with the tool name and arguments
3. The backend queues the search task and returns a task_id immediately
4. The client polls GET /v1/search/task/{task_id} with exponential backoff
5. When the task completes, results are returned to your agent

This architecture allows the backend AI agents to take the time they need for thorough research without blocking or timing out.

---

Environment Variables

Variable	Required	Default	Description
LEEROOPEDIA_API_KEY	Yes	—	Your Leeroopedia API key
LEEROOPEDIA_API_URL	No	https://api.leeroopedia.com	API endpoint
LEEROOPEDIA_POLL_MAX_WAIT	No	300	Max seconds to wait for a search task
LEEROOPEDIA_POLL_INTERVAL	No	0.5	Initial poll interval in seconds (grows via backoff)

---

Troubleshooting

"LEEROOPEDIA_API_KEY is required"

Set your API key in the MCP config:

{
  "mcpServers": {
    "leeroopedia": {
      "command": "leeroopedia-mcp",
      "env": {
        "LEEROOPEDIA_API_KEY": "kpsk_..."
      }
    }
  }
}

"Invalid or revoked API key" (401)

Double-check your API key at app.leeroopedia.com. Re-copy if needed.

"Insufficient credits" (402)

Purchase more credits at app.leeroopedia.com.

"Rate limit exceeded" (429)

Wait for the retry period before making more requests.

"Search timed out" (504)

The search task didn't complete within the poll window. Try a more specific query, or increase LEEROOPEDIA_POLL_MAX_WAIT.

---

Contributing

We welcome contributions! Please see our Contributing Guide for details on how to get started.

This project follows our Code of Conduct.

License

This project is licensed under the MIT License.