Semantic Compression API

Compress everything your AI reads.

Stop paying for tokens your AI doesn't need. Our semantic compression API cuts 85% of your context window costs — same meaning, fewer tokens, one API call.

Get Free API Key See How It Works

Powering context windows in

Claude Code

Cursor

Gemini CLI

Codex

Windsurf

VS Code

Compression Pipeline

Deterministic compression in 4 stages.

Every document follows the same reproducible path from raw text to compressed skeleton. No hallucinated summaries — only ranked graph nodes from your actual content.

Stage 01 // Ingestion

Document Analysis

Text chunked, analyzed, and scored semantically. Compression graph assembled.

Stage 02 // Ranking

PageRank Scoring

Graph edges weighted by semantic similarity. Importance propagated through the network.

Stage 03 // Extraction

Skeleton Generation

Top-ranked nodes form the compressed skeleton. Target ratio controls fidelity.

Stage 04 // Delivery

MCP Response

87.4% fewer tokens returned to your AI tool. Expandable on demand.

How It Works

Graph-based semantic compression.

Documents are chunked, embedded, and assembled into a semantic graph. PageRank scores identify the most important concepts. Only the highest-ranked nodes survive into the compressed skeleton — cutting 85-90% of tokens while preserving meaning.

Adaptive 3-tier compression engine
AST-aware code compression for 7+ languages
Multi-tenant scoping with workspace isolation

COMPRESSION

Semantic Graph

PageRank-based importance scoring

87.4% avg savings

Developer First

120+ MCP tools. One URL.

Drop into any AI tool that supports the Model Context Protocol. Claude Code, Cursor, Windsurf, VS Code — they all get the full compression layer with zero configuration beyond a config entry.

CWE-22 path traversal prevention on all file I/O
Async batch ingestion — 4x throughput with concurrency
Prometheus metrics, OpenTelemetry tracing, health checks

View API Reference

Connect Claude Code

mcp config

{
  "mcpServers": {
    "gotcontext": {
      "url": "https://api.gotcontext.ai/mcp",
      "headers": {
        "Authorization": "Bearer gc_your_key_here"
      }
    }
  }
}

terminal

live

# Install the MCP server

pip install semantic-modulator

# Start with stdio transport

python -m src.server

# Or use the MCP tool directly

> ingest_document(file_path="./docs/api.md")

> generate_skeleton(doc_id="api.md", ratio=0.15)

# Result: 485 → 61 tokens (87.4% reduction)

87.4%

Avg Token Savings

120+

MCP Tools

<90ms

P99 Latency

3,500+

Tests Passing

Try it now

Paste any text and see how much you can save. No signup required.

Input text

0/5,000 chars

Compressed output

Compressed text will appear here...

Pricing

Compress at your scale.

From solo developers to enterprise teams — pay only for what you compress.

Free

$0/month

Free forever

Get started without a credit card. Great for evaluation and small projects.

1,000 compressions/month
100KB max document
Standard compression
Community support

Get Started Free

Pro

Enterprise

Custom

Let's talk

Unlimited scale, self-hosted deployment, SSO, and a dedicated support team.

Unlimited compressions
Self-hosted option
SSO & SAML
SLA guarantee
Dedicated support

Contact Sales

By the numbers

Numbers that don't lie.

Every metric below is reproducible from the open-source codebase. No marketing copy — just measured results.

87.4%

Average token savings

Proven on real quantum computing documents

120+

MCP tools

Works with Claude, GPT, Gemini, Codex

3,500+

Tests passing

Across the compression engine

<90ms

Pipeline latency

Ingest — compress — return

“Dropped our Claude API bill by 70% in the first week. The MCP integration is seamless.”

Senior Engineer · AI Startup

“Deploy and forget. Connected once, saving tokens on every context window since.”

Tech Lead · Dev Tools Company

“The compression quality is impressive — our LLM outputs are indistinguishable from uncompressed.”

ML Engineer · Enterprise SaaS

Start today

Ready to compress?

Join AI developers who stopped burning tokens on redundant context and started compressing.

Start Compressing Read the Docs