Blocks wasteful tool calls before they execute. Filters bloated API responses on the wire. Gets smarter every session from its own mistakes.
Other tools compress what already happened. This one prevents waste before it executes, and tunes itself from real usage data.
Hooks fire before the tool call runs. A blocked read costs zero tokens. Other tools can only trim after the damage is done.
A local proxy strips SSE heartbeats, base64 images, and bloated tool results before they hit your context window.
When Claude retries a denied call, that's a false positive. The system detects it, reports it, and the model adjusts weights and thresholds automatically.
Most tools are static. claude-turbo has a feedback loop that tunes itself from aggregate usage data across all clients.
Deny a tool call (e.g., redundant file read)
Detect override — Claude retries same args within 60s? That's a false positive.
Record override count per heuristic in cumulative stats
Apply new model params — thresholds shift, weights change, bad heuristics get disabled
Aggregate fires + overrides across all clients
Compute FP rate per heuristic. >30% = penalize, <5% = boost.
Retune — adjust weights AND thresholds. High FP on H1? Allow 2 re-reads instead of 1.
Hook layer blocks waste upfront. Proxy layer filters the wire. Model layer tunes everything from evidence.
7 hook heuristics block waste upfront. 3 proxy filters clean the wire. All thresholds tunable from the server.
cargo test / pytest? Rewrites to scope to the active test.go build / cargo check twice? Denied. Output hasn't changed.--oneline -20. Limit tunable from server.ls -R / unbounded find? Pipes through head -50.bash_progress events dropped at the wire. Zero tokens.data:image/... payloads replaced with placeholder. Saves 10K+ tokens per image."Heavy refactoring session with Opus. claude-turbo blocked 47 redundant reads and 12 duplicate builds."
"Context window lasted 3x longer before /compact. The hook caught every re-read of my 800-line config file."
"Hit rate after a week. The model tuned H1 threshold to 2 for my workflow. Zero false positives since."
claude-turbo runs locally and intercepts tool calls on your own machine. The only thing that ever goes to our server is a handful of counters — and even that you can turn off in one line.
macos, linux, windows)Read// POST https://api.claude-turbo.xyz/v1/telemetry { "client_id": "a3f2c8d19e4b7f60c1d8e2a5f9b4c7e1", // random, local "version": "0.3.0", "platform": "macos", "days_active": 12, "total_tokens_saved": 184320, "total_interceptions": 247, "total_invocations": 1893, "heuristic_totals": { "H1_redundant_read": 94, "H4_duplicate_bash": 38, /* ... */ }, "heuristic_overrides": { "H1_redundant_read": 2, /* ... */ }, "daily": [ { "date": "2026-04-11", "tokens_saved": 12840, /* ... */ } ], "proxy_running": true }
~/.claude-turbo/config.json:
"telemetry_enabled": false
"telemetry_url": "https://your.host"
github.com/blasrodri/claude-turbo
Two commands. No config. Gets smarter every session.
cargo install claude-turbo && claude-turbo install