An MCP Server for Claude, Cursor & Windsurf

40+ Tools.
One MCP Server. Zero Waste.

Add Lemma as an MCP server and your agent gets shared semantic memory, zero-LLM code analysis, and surgical AST refactoring — no proxy, no baseURL changes, no rewriting your stack.

Pays for itself if it saves you ~$0.40/day · 7-day free trial · Cancel anytime

Works in Claude Code, Cursor & WindsurfSetup in 2 minutesCancel anytime
claude — mcp tools
mcp
$ claude mcp add lemma -- npx -y @nxuss/lemma mcp
+ lemma connected · 40 tools available
> "What breaks if I change engine.ts?"
tool: lemma.depgraph("src/core/engine.ts")
14 edges · 0 LLM calls · 100% local AST
$ > "Have we solved JWT validation before?"
tool: lemma.search_memory("JWT validation")
91% match · The Brain (shared team memory)
Saved $0.02 · 12ms · 0 cloud tokens
Every session prints its own receipt — see it below, not a headline number.
40+ MCP tools · zero-LLM analysis

Your Agent Already Has These Tools.

Free covers the cache. Pro unlocks 17 power tools — code intelligence, surgical AST edits, the squeezer, and PR review.

PRO

Dependency Graph

Blast-radius map of every import. Know what breaks before you refactor.

PRO

Coupling Radar

Circular deps, god modules, dead islands. The architecture smells, listed.

mcp:coupling_radar
Unlock with Pro →
PRO

Entropy Score

Complexity 0–100 per file. Refactor the worst offenders first.

mcp:entropy_score
Unlock with Pro →
PRO

Dead Export Necromancer

Zombie exports nobody imports — inflating context, wasting tokens.

mcp:dead_export
Unlock with Pro →
PRO

Pattern Fossil Hunter

Legacy patterns tracked: callbacks, var, require(). Measure modernization.

mcp:pattern_fossil
Unlock with Pro →
PRO

Precog Static Predictor

Entropy + coupling + git history predicts which files bug out next.

PRO

Git Heatmap Risk

Churn + co-editing patterns. Files that change together break together.

mcp:git_heatmap
Unlock with Pro →
PRO

Surgical AST Insert

Insert methods and types via the TS compiler. No regex, no strings.

mcp:ast_insert
Unlock with Pro →
PRO

Declarative Refactor

Rename or move across the workspace. Every import updates. Zero errors.

PRO

Smart Workspace Patch

Multi-file search-and-replace with duplicate protection.

mcp:smart_patch
Unlock with Pro →
PRO

Sandbox Validation

Patches validated in an isolated TS sandbox before they touch your code.

PRO

Auto-Heal

Reads the stack trace, finds the root cause, generates the fix.

mcp:auto_heal
Unlock with Pro →
FREE

Exact Match Cache

Instant dedup of identical prompts. <2ms, fully local.

mcp:cache_l1
FREE

Semantic Memory (CARS)

Same intent, different words → answered locally in <50ms.

mcp:cache_l2
PRO

Context Squeezer

Collapses repetitive logs and bloat before they reach the model.

PRO

Hive Mind Sync

A solves it once, B gets it free. Shared team cache in realtime.

mcp:hive_mind
Unlock with Pro →
PRO

Local Diff Review

Paste any diff: regressions, security, breaking changes. $0.00.

mcp:diff_review
Unlock with Pro →
PRO

PR Review

Score + verdict + findings on every PR. Auto-approve past threshold.

mcp:pr_review
Unlock with Pro →
PRO

CI/CD Generator

Emits the GitHub Actions / Azure workflow that runs review on every PR.

The Brain · mcp:search_memory

Permanent Semantic Memory.

Not a cache that invalidates. A knowledge base that grows. Your agent calls it over MCP — every solution your team discovers is stored locally and shared instantly across every developer and agent.

Free: 300 req/moPro: unlimited →
The Brain
Local knowledge base · Zero cloud cost
Try searching for "JWT validation" or "Prisma schema"
How It Works

Learn Once. Reuse Forever.

Every time you ask a question, Lemma checks The Brain first. If a semantically similar answer exists (≥75% match), it returns instantly from local storage — zero cloud cost, zero latency.

1. Query arrives
Lemma intercepts and vectorizes your prompt locally
2. Brain search
Semantic similarity scan against ChromaDB (&asymp;12ms)
3a. HIT (&ge;75%)
Response synthesized from local memory &rarr; $0.00
3b. MISS (<75%)
Forwarded to cloud LLM &rarr; response auto-cached in The Brain
Stored permanently, shared across your team
The Brain survives restarts, grows with every query, and syncs across developers via Hive Mind cloud. No more paying twice for the same answer.
Quick Wins

See Your Savings in 30 Seconds

Pick your profile. These are modeled from typical usage patterns, not a guarantee — your own receipt after week one is the real number.

👤
Solo Developer
$500/mo in AI costs
$325/mo saved immediately
65% of your current AI spend
Pays for itself in 5 days
Start Pro Trial →
Most Popular
Pro
🏢
Medium Startup(10 engineers)
$8,000/mo in AI costs
$5,200/mo saved immediately
65% of your current AI spend
Pays for itself in 1 week
Start Pro Trial →
🏛️
Enterprise(50+ engineers)
$50,000/mo in AI costs
$32,500/mo saved immediately
65% of your current AI spend
ROI in 2.6 hours of setup
Contact Sales →
Start Free Trial →

7-day free trial · Cancel anytime

Simple Pricing

Free Is a Demo. Pro Ships Your Product.

Start free. Upgrade when you’re saving money. $12/mo, locked for life.

Free vs Pro: What You Get

FeatureFreePro
Exact Match Cache
Basic Privacy Firewall
Semantic Cache (The Brain)
CARS Engine (Local Synthesis)
Context Squeezer
Code Intelligence Suite (7 tools)
Surgical AST Tools
Local Diff + PR Review
Telemetry Dashboard
Request Limit300/moUnlimited
Free
Standard Gateway
$0forever
Exact Match Cache (Layer 1)
Local Semantic Cache (Layer 2)
Privacy firewall (basic)
300 requests / month
Self-hosted local gateway
Pro tools, unlimited requests
Get Started Free

No credit card needed

Most Popular — Pro
Unlimited Local Power
$12/ month$29/mo
Early bird — locked-in for life
Everything in Free
Code Intelligence Suite (7 tools)
Surgical AST Tools (5 tools)
Context Squeezer & log collapsing
Local Diff + PR Review
CARS Engine & telemetry dashboard
Kiro Composer Bypass (smart patching)
Unlimited local requests

7-day free trial · Cancel anytime · Keep Free forever

Need managed private infrastructure? Talk to us about Enterprise →

Waiting on shared team memory (Cloud)? Join the Cloud waitlist

Licenses are per-instance and managed via lemma.nxus.studio. Questions? juancarlos@nxus.studio

Add as an MCP Server

One Command. 40+ Tools.

No proxy, no baseURL rewrite. Your agent connects to Lemma directly over MCP.

1

Connect Your Agent

claude mcp add lemma -- npx -y @nxuss/lemma mcp

Run this once in your project. Claude Code picks up the 40+ Lemma tools immediately.

2

Activate Pro (Optional)

lemma activate YOUR_LICENSE_KEY

Required for shared team memory (The Brain), semantic matching, and the CI/CD PR review tools.

3

Just Ask

No new syntax to learn. Ask your agent things like “check for circular dependencies in src/” or “have we solved this before?” — it calls depgraph, coupling_radar, or search_memory for you, automatically.

Cloud Early Access

Join the Cloud Early Access

Be among the first to use Lemma Cloud. Zero setup, managed infrastructure, and pay only for what you use.

We'll only send you updates about Lemma Cloud. Unsubscribe anytime.

$19

Starting Price

Cloud Starter plan

Zero

Setup Required

Plug & play in minutes

99.9%

Uptime SLA

Managed infrastructure

✓ Managed Infrastructure✓ Auto-Scaling✓ Pay As You Go
Objections, answered

Fair Questions.

Free covers the cache — exact + semantic match, 300 requests/month. Pro unlocks the 19 power tools that actually change how you work: the 7-tool Code Intelligence suite, 5 surgical AST editors, Context Squeezer, and local PR review — all unlimited. If Pro saves you ~$0.40/day in tokens, it pays for itself.

Stop paying twice
for the same answer.

19 Pro tools, unlimited local requests, and a memory that grows with every session. $12/mo, locked for life.

Cancel anytime · Keep Free forever

$0.40
/day to break even
100%
Self-Hosted
0
Vendor Lock-in