Stop runaway Claude API spend
in 60 seconds

Tourniquet is a free, local-first proxy that hard-caps your Anthropic API at a daily limit you set. When the cap hits, it kills the stream mid-response — your agent stops, your bill doesn't grow.

tourniquet — proxy running on :8787

$ claude "refactor everything in /src for performance" Streaming response via Tourniquet proxy... I'll start by auditing each module. Beginning with src/core/engine.py — the main event loop looks like it could benefit from async batching, and the inner loop at line 847 is allocating a new dict on every— ⚡ CAP HIT — $15.00/day limit reached Stream terminated. No further tokens billed. event: message_stop data: {"type":"message_stop","stop_reason":"tourniquet_cap_hit"}

Tourniquet dashboard — live spend, alert history, in-app one-tap recovery

This has already cost people real money

Three steps, then forget about it

Everything happens in the dashboard

Once Tourniquet is running, your browser opens at http://localhost:8787/dashboard. Every cap, every kill, every recovery — all from one screen. No config files to edit, no commands to memorise.

Get started now

Point your agent at it

Swap two env vars. Your agent code, SDKs, and prompts don't change.

export ANTHROPIC_BASE_URL=http://localhost:8787
export ANTHROPIC_API_KEY=tq_YOUR_TOKEN
claude

ANTHROPIC_BASE_URL=http://localhost:8787
ANTHROPIC_API_KEY=tq_YOUR_TOKEN

import anthropic

client = anthropic.Anthropic(
    api_key="tq_YOUR_TOKEN",
    base_url="http://localhost:8787",
)

resp = client.messages.create(
    model="claude-haiku-4-5-20251001",
    max_tokens=100,
    messages=[{"role": "user", "content": "hi"}],
)
print(resp.content[0].text)

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({
  apiKey: "tq_YOUR_TOKEN",
  baseURL: "http://localhost:8787",
});

const resp = await client.messages.create({
  model: "claude-haiku-4-5-20251001",
  max_tokens: 100,
  messages: [{ role: "user", content: "hi" }],
});
console.log(resp.content[0].text);

curl -s -N -X POST http://localhost:8787/v1/messages \
  -H "authorization: Bearer tq_YOUR_TOKEN" \
  -H "content-type: application/json" \
  -H "anthropic-version: 2023-06-01" \
  -d '{"model":"claude-haiku-4-5-20251001","max_tokens":50,"messages":[{"role":"user","content":"say hi in 5 words"}]}'

How Tourniquet compares

Comparison reflects the public state of each product as of May 2026. Corrections welcome via GitHub issue.

The cleanest stop in the category

When your daily cap is reached mid-stream, Tourniquet doesn't drop the TCP connection. It injects a synthetic SSE message_stop event at the end of the current partial chunk, then cleanly closes the response:

This matters because the stop_reason field is surfaced to your agent code. A well-written agent can catch tourniquet_cap_hit, log the event, and exit gracefully — rather than crashing with an unhandled network exception.

Common questions