bctx — Token Compression

Tracking since Apr 16, 2026.

Changelog

• May 27, 2026
  description

  better-ctx automatically compresses your prompts before they reach the AI — 
  saving up to 99% of tokens on every message.
  
  When you paste code, error logs, or long context into an AI chat tool, most 
  of it is noise: blank lines, comments, boilerplate, repeated patterns. 
  better-ctx strips it down to pure signal before the message is sent — 
  so the AI gets less clutter and you spend fewer tokens.
  
  HOW IT WORKS
  ────────────
  1. You type or paste your prompt as normal
  2. When you hit Send, better-ctx intercepts it
  3. Compresses the text using AST-aware patterns
  4. Shows you the savings: "1,200 → 180 tokens (-85%)"
  5. Sends the compressed message automatically
  
  NO CHANGES TO YOUR WORKFLOW
  ────────────────────────────
  Works silently in the background. No new UI to learn. 
  Just cheaper, faster AI interactions.
  
  IT SUPPORTED MOST OF SITES THAT PROVIDE LLM CHAT
  ───────────────
  
  FEATURES
  ────────
  ✓ Auto-compress on send (Enter key)
  ✓ Auto-compress on paste
  ✓ Token savings badge shows before/after count
  ✓ Works with full better-ctx binary for maximum compression
  ✓ Basic compression works even without binary installed
  ✓ Toggle on/off from the popup
  ✓ Zero telemetry — your code never leaves your machine
  
  FULL COMPRESSION (OPTIONAL)
  ────────────────────────────
  Install the better-ctx binary for AST-level compression across 18 
  programming languages — achieving up to 99% token reduction using 
  tree-sitter signatures, delta reads, and 90+ CLI output patterns.
  
  Install: npm install -g better-ctx-bin
  Or: cargo install better-ctx
  
  PRIVACY
  ───────
  This extension does not send your code to any server. 
  Compression happens locally — either in the browser or via the 
  locally installed binary on your machine.

  Bctx automatically compresses your prompts before they reach the AI —
  saving up to 90% of tokens on every paste and send.
  
  When you paste code, logs, or long context into an AI chat, most of it
  is noise: blank lines, comments, boilerplate, repeated patterns. bctx
  strips it to pure signal before the message is sent — less clutter for
  the model, fewer tokens for you.
  
  HOW IT WORKS
  ────────────
  1. Paste or type your prompt as normal
  2. Hit Send (or paste) — bctx intercepts it
  3. Compresses the text using context-aware patterns
  4. Shows savings in a small badge: "1,200 → 180 tok (-85%)"
  5. Sends the compressed message automatically
  
  SUPPORTED SITES
  ───────────────
  Ii is supporting almost every AI Chat like websites and work also on new one upon allowing the extension to do so.
  
  FEATURES
  ────────
  ✓ Auto-compress on paste (Cmd/Ctrl+V)
  ✓ Auto-compress on send (Enter key or Send button)
  ✓ Token savings badge shows before/after count and compression path
  ✓ Hold Shift while pasting to bypass compression
  ✓ Works with the bctx CLI for maximum offline compression
  ✓ Cloud fallback works immediately — no setup required
  ✓ Toggle on/off from the popup
  
  COMPRESSION PIPELINE
  ────────────────────
  The extension tries three paths in order:
  
  1. Local bctx CLI (optional install) — fastest, fully offline,
     AST-aware across 18 languages, up to 99% reduction
  2. Cloud API at api.betterctx.com — instant fallback, no setup
  3. Inline regex — always available, no network required
  
  OPTIONAL: INSTALL THE bctx CLI
  ───────────────────────────────
  For offline, maximum-quality compression:
    - brew install bctx
    - npm install -g bctx-bin
    - curl -fsSL https://bctx.dev/install.sh | sh
  
  PRIVACY
  ───────
  When the bctx CLI is installed: compression runs entirely on your
  machine. No text leaves your device.
  
  When using the cloud fallback (default): your text is sent to
  api.betterctx.com/compress for compression. The endpoint does not
  store your content. You can disable cloud compression in the popup
  or install the CLI to stay fully local.
  
  Full privacy policy: https://betterctx.com/privacy

• May 27, 2026
  short_description

  Compress code context before pasting into AI chat tools. Reduces token consumption by up to 99%.

  Compress pasted content before sending to AI tools. Reduces token consumption by up to 90% on paste and send.

• May 27, 2026
  name

  better-ctx — Token Compression

  bctx — Token Compression

• May 27, 2026
  host_permissions

  https://chatgpt.com/*, https://chat.openai.com/*, https://claude.ai/*, https://gemini.google.com/*, https://github.com/copilot/*, https://lovable.dev/*, https://*.bolt.new/*, https://bolt.new/*, https://v0.dev/*, https://poe.com/*, https://aistudio.google.com/*, https://labs.perplexity.ai/*

  https://chatgpt.com/*, https://chat.openai.com/*, https://claude.ai/*, https://gemini.google.com/*, https://github.com/copilot/*, https://lovable.dev/*, https://*.bolt.new/*, https://bolt.new/*, https://v0.dev/*, https://poe.com/*, https://aistudio.google.com/*, https://labs.perplexity.ai/*, https://www.perplexity.ai/*, https://grok.com/*, https://copilot.microsoft.com/*, https://betterctx.com/*, https://api.betterctx.com/*

Permissions & access

Permissions

storagenativeMessaging

Host access

https://chatgpt.com/*, https://chat.openai.com/*, https://claude.ai/*, https://gemini.google.com/*, https://github.com/copilot/*, https://lovable.dev/*, https://*.bolt.new/*, https://bolt.new/*, https://v0.dev/*, https://poe.com/*, https://aistudio.google.com/*, https://labs.perplexity.ai/*, https://www.perplexity.ai/*, https://grok.com/*, https://copilot.microsoft.com/*, https://betterctx.com/*, https://api.betterctx.com/*

Screenshots