ReplayCI

LLM function calling reliability platform. Validate, monitor, and gate tool-call behavior across providers.

• Contract testing — define expected tool-call behavior in YAML, test against any LLM provider
• Provider-agnostic — same contracts work with OpenAI, Anthropic, and more
• Failure classification — every failure gets a taxonomy label and a stable fingerprint
• Determinism proof — run contracts multiple times to verify response stability
• Dashboard — push results to app.replayci.com for tracking and history

Quick start

npm install -D @replayci/cli
npx replayci init

Set your API keys:

export REPLAYCI_PROVIDER_KEY="sk-..."    # OpenAI or Anthropic key
export REPLAYCI_API_KEY="rci_live_..."   # Get one at app.replayci.com/signup

Run:

npx replayci

Documentation

• Quickstart
• Writing Tests
• CLI Reference
• Security

License

ISC