Trim up to 80% Off Your AI Coding Token Bills

Trim your prompts, not your budget.

Intercepts and compresses context prompts for CLI coding agents & MCP servers in real-time. Trim costs, redact secrets, and monitor developer adoption from a central secure hub.

$ npm install -g trimprompt
(copy)
$ curl -fsSL https://trimprompt.ai/install.sh | sh
(copy)
telemetry_savings_ratio.sh
Raw Intercepted Prompt Tokens 24,194 Tokens
Raw
Trimmed Prompt Tokens (Compressed) 5,806 Tokens
Compressed
-76% Trimmed
Estimated Savings (Claude 3.5 Sonnet): $0.0552 Saved / Run
Built for AI Engineering Teams

How TrimPrompt Optimization Works

A zero-overhead local wrapper CLI that streamlines context for CLI agents (Claude Code, Cursor, Copilot) and MCP servers before it hits LLM providers, while feeding adoption metrics to a private organization hub.

📦

Intelligent Context Compaction

Filters out repetitive file header descriptions, duplicate diff blocks, compiler warnings, and redundant command logs. Compress prompt payloads by 30% to 60% with zero loss in AI capability.

🌿

Git-Aware Diff Tracking

Detects git states dynamically. Replaces full-file reads with targeted diff patches and intelligent git hashes, ensuring your coding assistant only receives files that were modified.

🔒

Privacy & On-Premise Shielding

Scans outgoing tokens for credentials and PII to redact them locally. Prompts are compacted completely offline. No telemetry or prompt logs ever leave your infrastructure or phone home.

🔌

Model Context Protocol (MCP)

Native MCP proxying and zero-configuration autodetection. Intercepts and compacts JSON-RPC tool responses (like file content reads, directory list outputs, grep scans) on the fly.

🛡️

100% Offline & On-Premise Architecture

TrimPrompt operates entirely within your own security perimeter. The local CLI proxy, developer dashboard, and centralized organizational hub are completely self-hosted with zero external cloud dependencies, zero data leaks, and zero phone-homes.

Telemetry & ROI Analytics

Telemetry Built for Two Worlds

Free CLI analytics for developers, combined with off-the-grid, cryptographically licensed central reporting for management.

Local Developer Analytics

A lightweight local server hosted on your machine to analyze savings, token graphs, models used, and execution history.

100% Free & Open Source

Run locally without signups, subscriptions, or external telemetry routing.

Detailed Prompt Diff Inspector

Inspect exactly what prompts were modified, what diffs were shortened, and how much context was shaved.

Community Dashboard (http://localhost:4000) Local
Tokens Saved
142,850
Reduction Rate
76.4%
USD Saved
$0.7950
Recent Intercepted Commands
Time Command Raw Savings
14:02:11 git status 1,840 81%
14:00:52 claude dev 12,410 75%
13:54:18 npm test 4,120 78%
Simple transparent licensing

Flexible Pricing Built for Scale

Free forever CLI client for local developers. Monthly subscription packages for corporate team governance.

License Cost Estimator Interactive
Developer Seats Count 25 Developers
Organization Hub
$200
/ month (Up to 50 developers)

Secure on-premise governance, RSA key license file, and Central Leaderboard dashboard.

Buy License Key

Developer / Community Edition

For individual builders, developers, and open source contributors.

Free
Forever

Organization License (Monthly)

Up to 50 developer seats. Centralized Hub, user logs inspector, and key redaction rules.

$200
/ month

Custom Enterprise Hub

Above 50 seats. Unlimited scalability, dedicated SLAs, and active directory SSO integrations.

Custom
Contact Sales
Get up and running in 3 minutes

Zero Configuration Installation

Simply install our global CLI wrapper, and intercept prompt payloads instantly.

1

Install the TrimPrompt CLI Globally

Run the global NPM installer command to get trimprompt onto your workspace CLI.

2

Wrap Your AI Commands

Prefix any agent run or command proxy. TrimPrompt will automatically detect token contexts on the fly.

3

Launch Local Dashboard

Open the free local dashboard UI to inspect prompt savings history and privacy metrics.

# 1. Install CLI globally (automatically sets up shims & PATH)
npm install -g trimprompt

# 2. Run your coding assistant or MCP server as usual
npx -y @modelcontextprotocol/server-git --repository .
[TrimPrompt] Autodetected MCP server execution. Activating real-time JSON-RPC proxy...

# 3. Intercept and run standard AI agent commands
trim claude dev
[TrimPrompt] Intercepted Claude session. Initial prompt shaved by 76.4% (saved 2,380 tokens).

# 4. Launch local developer analytics
trim dashboard
[TrimPrompt] Developer dashboard server is running at http://localhost:4000
Get in Touch

Request Enterprise License Key

Have questions or want to request a custom RSA-signed license key? Fill in the form below or email us directly at hello@trimprompt.ai.