Hit your Claude Code limit? Keep coding.
Edgee compresses tokens before they reach Claude. Same plan, same code, up to 2x more session time. Install in 30 seconds.
Works with Claude Code, Codex, OpenCode. Cursor coming soon.
$curl -fsSL https://edgee.ai/install.sh | bash
$edgee launch claude
✓Claude Code running with Edgee compression
Backed by founders and leaders of
You're paying for Max. You're still hitting limits.
The Claude Code usage limit resets every 5 hours, and the weekly Opus ceiling can lock a Max plan mid-sprint.
5-hour window
You burn through your quota before lunch and wait until evening to resume.
Weekly Opus limit
One deep session and Opus is locked for 6 days. You fall back to Sonnet for the rest of the sprint.
Pay-per-token overage
Extra usage kicks in and your monthly cost doubles without warning.
One command. Every token, compressed.
Extend your Claude Code session with edge-native token compression, no config changes, no proxy setup.
# Install
curl -fsSL https://edgee.ai/install.sh | bash
# Launch Claude Code with compression enabled
edgee launch claude
# That's it. Compression runs silently.
# Full observability in your Edgee dashboard.
Up to 80% tool compression
Tool results are compressed at the edge before reaching Claude, without losing context quality.
Zero code changes
Edgee wraps your existing Claude Code install. Your configs, MCP servers, and CLAUDE.md stay untouched.
Full observability
Track tokens saved, cost avoided, and latency per session in your dashboard.
With Edgee vs without
What the Claude Code weekly limit looks like before and after.
| Claude Code alone | Claude Code + Edgee | |
|---|---|---|
| Session duration before limit | ~2h | ~3–4h |
| Weekly Opus capacity | Locked fast | Stretched |
| Token cost per session | Full | Up to 50% less |
| Setup time | — | 30 seconds |
Claude Code with and without Edgee token compression.
How the compression works
- Your Claude Code
- Edgee compresses at the edge
- Claude API
Edgee sits between your coding agent and Claude. Prompts are compressed at the edge before being forwarded, then decompressed server-side for the model. Output quality stays identical.
Questions devs actually ask
Does it work with my existing CLAUDE.md and MCP servers?
Will the compression affect Claude's output quality?
What happens if Edgee is down?
Is my code sent to Edgee servers?
Does it work on Windows / WSL?
What's on the roadmap for Cursor and OpenClaw?
Keep coding
Stop watching your limit reset timer.
No credit card. Works with your existing Anthropic plan.