Record, Inspect, Replay AI Agents
The agent in your app stack went off the rails again. Stop digging through logs. Orchid records the full run. Inspect and replay it locally, step by step.
Spin up the local demo
Once running, open http://localhost:4321 in your browser and enter the API key orchid_live_8675309 to explore the pre-seeded demo trace.
Meet Orchid, the Orchestration interactive debugger
Stop digging through logs when agents go off the rails. Orchid records every API call for step-by-step local inspection & replay.
Records Any API
OpenAI, Anthropic, and Gemini recognized natively. Any other HTTP API, search, vector stores, internal services, captured with one header.
MCP Native
AI assistants query your traces directly. Search payloads, triage failures, export fixtures — all through Model Context Protocol.
Deterministic Replay
Blocks outbound network traffic and serves recorded session fixtures during local development and testing. 100% offline, deterministic, and free.
Real-Time Cost Tracking
Every exchange gets USD cost attribution from a pricing engine you control. Roll-ups per session and execution step.
Web Visualizer
Interactive pipeline graph UI. Time-travel through events, inspect payloads, view latency profiles — all in the browser.
Local-First by Design
Single binary, local SQLite, no cloud backend, no phone-home. Secret-like headers and fields are redacted before anything touches disk.
Debugging AI shouldn't feel like archaeology.
When your LLM is buried deep inside a backend service, finding the root cause of an unexpected output means searching through logs and guessing what was sent over the wire.
Stop sifting through terminal dust.
Orchid captures the exact artifact. Route your traffic through the proxy, and instantly inspect every raw prompt, schema, and response in high-fidelity, without altering a single line of your core business logic.
The best part - everything is stored locally, no data ever leaves your infrastructure.
How it works
Proxy Traffic
Import the Orchid SDK. Point your app to the proxy and run in capture mode mode. Standard libraries route natively — zero code rewrites.
Store & Inspect
Every LLM request/response is captured into a local thread-safe SQLite database. Inspection runs via the Web Visualizer or MCP queries.
Replay & Test
Export portable JSON fixtures. Run offline test suites with replay mode — zero API cost, fully deterministic.
Deterministic Testing
Kill flaky tests with replay mode
In replay mode, the proxy blocks all outbound traffic. Requests are matched against recorded responses by semantic hash and served instantly. No API calls, no costs, fully deterministic.
Without Orchid
- →Tests hit live APIs — slow, expensive, rate-limited
- →Non-deterministic outputs cause flaky assertions
- →Manual mock maintenance per model update
- →CI costs scale linearly with test count
With Orchid
- →Saved fixtures serve responses locally instantly
- →Identical response every run, fully deterministic
- →Re-record fixtures after making agent changes
- →CI LLM spend: $0.00
Runs Local. Stays Local.
Orchid runs entirely inside your infrastructure. No telemetry, no external cloud.
The Architecture
Local-First
- 100% Offline: Run via Docker on your local machine or private cloud.
- Zero Data Leaks: SQLite session recordings never leave your network.
- Open SDKs: Intercept and replay at the transport layer with zero vendor lock-in.
The Builder
Mario Guerra
- 18 years shipping developer tools and AI platforms.
- Drove TypeSpec adoption across 30% of Azure at Microsoft.
- Onboarded 150+ customers to AI hardware at Qualcomm.
Ready to debug smarter?
Start recording and replaying your AI pipeline traffic locally in minutes.