OrcaReplay

What is OrcaReplay?

OrcaReplay records what a coding agent does, so you can inspect, repeat and compare a run afterwards instead of guessing why it failed. It installs as the orcareplay npm package and adds one command, orca. A local proxy sits in front of the model API, with extra layers for shell exit codes, timing, file changes and MCP traffic, so your agent needs no changes beyond two environment variables. The recording is a file: orca show lists every model turn, tool call and file change, orca replay serves the same run back byte for byte with network access blocked and no tokens spent, and orca compare forks it at a chosen checkpoint onto several models and grades each with a command such as npm test. The orca ui command opens a self-contained HTML timeline. It comes from the team behind OrcaRouter, whose gateway is the default for model comparisons, and any OpenAI-compatible gateway works. The project is open source and free to install.

Use cases

• Record a coding agent run for later debugging
• Replay an agent session offline for free
• Compare models from the same checkpoint
• Find which tool call changed a file
• Share an agent bug report as an HTML timeline
• Capture an agent's system prompt

Who it's for

• Developers using coding agents
• Agent harness builders
• AI engineers
• Open-source maintainers

Why use OrcaReplay

• Turns a failed agent run into a file you can reopen later, so you debug evidence instead of scrolling a terminal.
• Replays cost nothing because the network is blocked and no model is called.
• Compares models fairly: files and conversation prefix stay the same and only the model changes.
• Needs no changes to your agent because capture happens at the process and socket boundary.
• Exports a single self-contained HTML file you can attach to a bug report.

OrcaReplay features

• orca record wraps an unmodified agent with a local proxy
• Byte-for-byte replay with network egress blocked
• Fork from any checkpoint onto a different model
• orca compare table of verdict, tokens, cost and wall time
• Capture of shell exit codes, timing and file changes
• Event graph linking tool calls to their effects
• Self-contained HTML timeline via orca ui or orca export
• SVG, PNG and GIF bug cards for issues
• orca quickstart demo that needs no key or tokens
• Opt-in TLS interception for agents that cannot be redirected

OrcaReplay integrations

• Claude Code
• Codex CLI
• OrcaRouter
• Claude
• GPT
• Gemini
• Grok
• DeepSeek
• Qwen
• Ollama

How to get started with OrcaReplay

1. Install the package with npm i -g orcareplay, which puts the orca command on your PATH.
2. Run orca record followed by your agent, such as claude, and let it work as usual.
3. Inspect the run with orca show, orca graph or the HTML timeline from orca ui.
4. Run orca replay last to reproduce it with no network and no tokens.
5. Fork with orca replay or orca compare using --from and --models, and grade results with a verify command.

OrcaReplay limitations

• Recording and forking need an installed agent, a key, a network and real tokens
• Raster and GIF export needs a browser toolchain such as playwright-core installed separately
• File changes are snapshotted once per turn, so linking a change to one tool call is inferred

OrcaReplay pricing

OrcaReplay is open source, so you can use and self-host it for free.

Prices can change and may vary by country or exclude tax. Check the official pricing page.

OrcaReplay alternatives

The main OrcaReplay alternatives listed on Debuly are Langfuse, Helicone, Opik, Braintrust, IronCurtain and Arize Phoenix. The table compares pricing and platforms; order is A–Z.

1. 

   Arize Phoenix

   Open-source tool for tracing, evaluating and experimenting on LLM apps and agents, free to self-host or run locally.

   Developer platformsAutomation and agents
   Open sourceVisit Arize Phoenix (official site)
2. 

   Braintrust

   AI observability and evaluation platform for tracing LLM apps and agents, scoring outputs and catching regressions before release.

   Developer platformsCoding
   FreemiumVisit Braintrust (official site)
3. 

   Helicone

   Open-source LLM observability and AI gateway that logs requests, tracks cost and latency, and routes to 100+ models through one API.

   Developer platformsCoding
   FreemiumVisit Helicone (official site)
4. 

   IronCurtain

   Open-source runtime that limits what AI agents can do, using security policy compiled from a plain-English constitution.

   Developer platformsAutomation and agentsCoding
   Open sourceVisit IronCurtain (official site)
5. 

   Langfuse

   Open-source LLM observability and evaluation platform for tracing agents, managing prompts and scoring outputs.

   Developer platformsAutomation and agentsCoding
   FreemiumVisit Langfuse (official site)
6. 

   Opik

   Comet's open-source platform for tracing, evaluating and monitoring LLM applications and agents, self-hosted or in the cloud.

   Developer platformsAutomation and agentsCoding
   FreemiumVisit Opik (official site)

OrcaReplay FAQ

Sources & verification

Facts on this page come from the official sources below, each with the date we read it. Written by Debuly; the vendor did not review this page.

Image credits

• Logo: from avatars.githubusercontent.com · fetched 24 Oct 2026 · Vendor trademark, used only to identify the product; removed on request
• Cover: none collected yet, so a generated tile is shown.

Spot an error? Report an issue with this listing. We also remove a logo or image on the owner's request.