Quickstart

Get an API key and compress your first prompt with the hosted API, in under 5 minutes.

Nothing to install The hosted API does the compression for you. Get an API key and call it from any language.

1. Get an API key

Head to the dashboard, sign in and create a key. Keys look like sc_live_xxxxxxxx.

Set your key as an environment variable:

export SUPERCOMPRESS_API_KEY=sc_live_YOUR_KEY

2. Compress your first prompt

Send your context and your question to the hosted API. Here the context is an account summary, and the query asks about the current plan:

curl -X POST https://www.supercompress.dev/api/v1/compress \
  -H "X-API-Key: sc_live_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "context": "## User account info\nUser ID: 4521\nPlan: Pro\nBilling cycle: monthly\nSeats: 5\n\n## Feature requests (Q2)\nMulti-workspace view, in progress\nDark mode, planned\nAPI rate limit increase, under review\n\n## Recent activity\n2026-06-28, upgraded from Starter to Pro\n2026-06-25, generated 3 API keys\n2026-06-20, hit rate limit, filed support ticket",
    "query": "What plan is the user on and how many seats do they have?"
  }'

Prefer a one-liner? A form-encoded POST to /compress works with any HTTP client:

curl -d "context=Your long context here...&query=What do you want to know?" \
  https://www.supercompress.dev/compress \
  -H "X-API-Key: sc_live_YOUR_KEY"

3. Understand the response

Example response:

{
  "compressed_text": "…",
  "original_tokens": 104,
  "kept_tokens": 40,
  "tokens_saved_pct": 61.5,
  "important_kept_pct": 0.95,
  "compression_risk": "low",
  "kept_blocks": ["User account info", "Recent activity"],
  "dropped_blocks": ["Feature requests (Q2)"]
}

Notice that the Feature requests section appears in dropped_blocks: it was dropped because it is irrelevant to a question about the current plan. The sections that answer the question were kept.

  • compressed_text: the compressed context to send to your model.
  • original_tokens and kept_tokens: token counts before and after compression.
  • tokens_saved_pct: the share of tokens removed.
  • compression_risk: a risk level for this compression, such as low.
  • kept_blocks and dropped_blocks: which sections of the context were kept or dropped.

4. Use it in a coding agent

One npm command wires SuperCompress into your coding agent:

npm i -g supercompress-proxy && supercompress setup

Then check the setup:

supercompress doctor

See the coding agents guide for per-agent instructions.

5. Use it from Python

The supercompress package ships a thin Python client for the hosted API:

pip install supercompress

Then, with the same context and question as above:

from supercompress.client import SuperCompress

sc = SuperCompress()  # reads SUPERCOMPRESS_API_KEY
out = sc.compress(context, question)
print(out.compressed_text)

6. Next steps

  • Read the API reference: POST /api/v1/compress (or just /compress), response types, and client configuration. See the API reference.
  • Wire it into your app, add compression before OpenAI, LangChain, Express.js, or any LLM call. See integrations.
  • Try the playground, test compression on your own text in-browser at supercompress.dev/playground.

The Python package also includes an offline function for local experiments; see the API reference.