Skip to main content
Use AI for more of your work. Keep your standards. Connect your repo and EvalGate finds the AI in your code: model calls, prompts, agents, and tools. Test what they do, catch regressions before release, and keep the proof. CI for AI behavior: trace → eval → gate. Run the public refund demo to reproduce PASS → BLOCK → PASS with the published SDK. Independent clean-install evidence. Agents can start with the structured JSON API view or the agent discovery guide. The public Skills repository includes AGENTS.md, the Agent Plugins manifest and bundled skills/MCP configuration. Try the credential-free API test mode with synthetic text. The product is one operating loop: trace to eval to gate. Real behavior produces evidence, reviewed failures become test cases, and promoted test cases become CI gates before a prompt, model, retriever, or agent change ships.

Quick start

Run a reviewed case, accept a passing baseline, and watch a deliberate regression exit 2.

Trace to eval to gate

Understand the operating loop before you wire in platform features.

SDK and CLI

Install the TypeScript or Python SDK and use the same assertions locally, in app code, and in CI.

Framework recipes

Copy setup paths for Node, Next.js, Python, FastAPI, LangChain, and RAG apps.

API reference

Integrate directly with the EvalGate platform for traces, runs, projects, and keys.

Feature status

See which surfaces are Shipped (live), which client gaps remain, and where to find verification evidence.

Platform map

Task-oriented map of EvalGate surfaces when the product feels too large.

Evaluation layers

How application evaluation, evaluator measurement, qualification, and authorization fit together.

​
The adoption path

1

Prove one local gate

Install the SDK without an account, replace the placeholder with a real check, and accept a reviewed passing baseline. Run the same gate after a deliberate regression before adding it to CI.
2

Capture real failures

Add tracing when local gates are not enough. EvalGate captures production and staging behavior with inputs, outputs, tool calls, latency, token usage, cost, and metadata.
3

Promote failures into coverage

Convert repeated or high-risk failures into reusable eval cases. Label, cluster, synthesize, and review cases so your suites track actual user pain.
4

Gate releases with evidence

Run the eval suite in CI, compare against the baseline, and give reviewers clear pass/fail evidence before changes merge.

​
What to use first

EvalGate surfaces marked Shipped in Feature status are live and available. Review that inventory for client gaps before depending on a convenience SDK or CLI method.
Judges and other model-backed workflows are bring your own provider key (BYOK): your provider or gateway credential is used and bills model usage directly. Understand BYOK and model-provider setup.

​
Explore next

CI/CD integration

Wire EvalGate into GitHub Actions or GitLab CI to gate every PR.

Tracing setup

Capture the real AI behavior that should become eval coverage.

LLM judge

Add judge-backed scoring when assertions alone are not enough.

Model providers and BYOK

Connect your provider or gateway credential and understand the separate authentication, billing, and data boundaries.

Agent governance

Scale from one gate to governed AI release workflows.