Which AI do you use?

ContextForge wraps your AI to add context optimization — drop any file and see how many tokens you save.

Your ContextForge API key

Stored locally in your browser and sent only with your requests.

No key yet? Type to try free — or get your own key.

Authorize from Claude

Copy your ContextForge MCP URL

This is the URL Claude will use to call ContextForge tools.

Open Claude → Settings → Integrations

In claude.ai, go to your account settings. Click Integrations, then Add integration. Paste the URL above and click Save.

Open Claude Settings →

Test the connection in Claude

In any Claude conversation, type /tools or ask Claude to "list ContextForge tools". If it shows the tools list, you're connected.

Connected to Claude — ContextForge tools are active in your Claude conversations.

Authorize with OpenAI

Enter your OpenAI API key. It’s validated directly with OpenAI, stored locally in your browser, and unlocks GPT models in ContextForge Chat — sent only alongside your chat requests, never stored or logged on our servers.

OpenAI key saved — GPT models are now unlocked in ContextForge Chat.

Authorize with NVIDIA NIM

Enter your free NVIDIA API key to use Nemotron. Get yours free at build.nvidia.com. Stored locally.

NVIDIA key saved — using Nemotron for free AI requests.

Connect local Ollama

Enter your Ollama base URL (default: localhost:11434). ContextForge will test the connection and list available models.

Ollama connected — using local LLM.

Compress your file

Drop any file — your AI gets smarter answers at a fraction of the token cost.

Drop your file here

e.g. sales-report.csv · notes.pdf · transcript.mp3

PDFWord ExcelCSV PowerPointImages AudioVideo CodeText

Compressing

Something went wrong

Compressed

—

Original tokens

—

After compression

—

Tokens saved

Compressed output — ready to paste into Claude or ChatGPT

Data Insights

What columns does this have?

How many rows?

Show me the first 10 rows

How many missing values?

Get your own API key