ShipGauge
Pre-ship gate for indie AI

Ship AI features without surprise bills or silent regressions

Cost guardrails + prompt eval in one CI check. No prompts stored.

shipgauge-bot commented on PR #42
Gate passed

Pass rate: 97% (baseline 96%)

Avg cost: $0.0021 / request

Gate blocked

Pass rate: 88% (▼ 8% vs baseline)

Avg cost: $0.0028 (+33%)

Cost guardrails + prompt eval in one CI check
Built for indie AI teams shipping fast

Three problems, one gate

ShipGauge tells you if you should ship — not just what you spent.

Bill shock

A dozen AI features, one giant invoice. You can't tell which feature burned the budget.

Silent regression

A prompt tweak drops quality 8%. Nobody notices until users complain three days later.

Tool sprawl

Langfuse is heavy, Promptfoo has no prod budget, LLMeter has no eval. You install three tools or give up.

How it works

SDK → CI → Production guard

01

Tag with SDK

Add feature_id to every LLM call with 5 lines of code. See per-feature spend in minutes.

02

Gate in CI

Run smoke evals on every PR. Block merges when pass rate drops or cost spikes.

03

Guard production

Set per-feature budgets. Soft warn at 80%, hard block at 100%. No runaway agents.

Why ShipGauge vs the rest

Promptfoo tells you if it's worse. LLMeter tells you if it's expensive. ShipGauge tells you if you should ship.

Per-feature cost

ShipGauge
LLMeter
Promptfoo
Langfuse

Production budget hard block

ShipGauge
LLMeter
Promptfoo
Langfuse

CI eval gate

ShipGauge
LLMeter
Promptfoo
Langfuse

Cost scorer

ShipGauge
LLMeter
Promptfoo
Langfuse

No prompt storage

ShipGauge
LLMeter
Promptfoo
Langfuse

Indie flat pricing

ShipGauge
LLMeter
Promptfoo
Langfuse

Simple indie pricing

Flat rate. No per-seat surprises. Join waitlist for early access.

View full pricing →
Bundle

StackLedger + ShipGauge Pro

Pro limits on both products — unified burn + pre-ship gate

$39/mo

Hobby

$0

For side projects and early experiments

  • 1 project
  • 10k events / month
  • 7-day data retention
  • 3 budget rules
  • GitHub Gate
Most popular

Pro

$29/mo

For indie AI SaaS shipping to production

  • 3 projects
  • 100k events / month
  • 90-day retention + anomaly detection
  • Unlimited budget rules
  • Slack alerts + CSV export

Team

$79/mo

For small teams with multiple AI features

  • 500k events / month
  • 1-year retention
  • 5 team seats
  • Priority support (24h)

Frequently asked questions

No. ShipGauge only stores metadata: model, tokens, cost, latency, and feature_id. We never store prompt or response text. This is a core architectural constraint, not a toggle.