# Oqoqo

## Docs

- [Core Concepts: Experiments, Trials, Rubrics, and Traces](https://docs.oqoqo.ai/core-concepts.md): Understand the key building blocks of an Oqoqo experiment: tasks, rubrics, sandboxes, trials, agents, treatments, traces, and metrics.
- [Frictions: Find Where Agents Struggle on Your Product](https://docs.oqoqo.ai/evaluation/frictions.md): Frictions are the steps where an agent stalled, retried, or gave up. Oqoqo surfaces them automatically so you know exactly where to improve.
- [Metrics: Compare Agent Performance Across Trials and Runs](https://docs.oqoqo.ai/evaluation/metrics.md): Oqoqo captures steps, tokens, cost, duration, and tool call breakdowns for every trial. Compare metrics across agents, treatments, and versions.
- [Rubrics: Define Evaluation Criteria for Experiments](https://docs.oqoqo.ai/evaluation/rubrics.md): Define pass/fail criteria for your Oqoqo experiments. Oqoqo rewrites your rubric to eval best practices and scores every trial automatically.
- [Traces: Inspect Every Step of an Agent Run in Oqoqo](https://docs.oqoqo.ai/evaluation/traces.md): Every agent trial is recorded as a trace: tool calls, file reads, errors, and reasoning steps — so you know exactly where things went wrong.
- [Agents and Models: Configure Who Runs Your Experiments](https://docs.oqoqo.ai/experiments/agents-and-models.md): Choose which AI coding agents and model variants run your Oqoqo experiments. Compare Claude Code, Codex, and more across effort levels.
- [Defining an Oqoqo Experiment: Tasks, Rubrics, and Library](https://docs.oqoqo.ai/experiments/defining-experiments.md): Learn how to define a repeatable Oqoqo experiment: write a task, add a rubric, attach your library, and select agents and treatments.
- [Experiments Overview: Run Agent Workflows at Scale](https://docs.oqoqo.ai/experiments/overview.md): An Oqoqo experiment bundles a task, agents, treatments, and environment into one repeatable unit you can run, compare, and iterate on.
- [Sandboxes: Isolated, Reproducible Environments for Trials](https://docs.oqoqo.ai/experiments/sandboxes.md): Oqoqo runs each trial in a clean, isolated sandbox with real repos, data, and files — so every run is reproducible and comparable.
- [Treatments: Test Variables Across Agent Experiments](https://docs.oqoqo.ai/experiments/treatments.md): Use treatments to isolate variables in your Oqoqo experiments: compare raw agent behavior against skills, MCP servers, updated APIs, and more.
- [What Is Oqoqo: Agentic Experimentation from Run to Result](https://docs.oqoqo.ai/introduction.md): Learn how Oqoqo helps teams run repeatable agent experiments in production-like environments and measure the effect of every change.
- [Oqoqo Quickstart: Define and Launch Your First Experiment](https://docs.oqoqo.ai/quickstart.md): Set up your first agent experiment on Oqoqo: join the waitlist, define a workflow, configure agents and treatments, and launch your first trial.
- [Frequently Asked Questions About the Oqoqo Platform](https://docs.oqoqo.ai/reference/faq.md): Answers to common questions about Oqoqo: free trials, setup time, supported agents, how experiments differ from sandboxes, and more.
- [Oqoqo Supported Agents: Claude Code, Codex, and More](https://docs.oqoqo.ai/reference/supported-agents.md): Oqoqo supports Claude Code and Codex today, with Cursor, GitHub Copilot, Antigravity, and OpenCode coming soon. Configure model and effort per agent.
- [Agent-Testable Surfaces: SDKs, APIs, CLIs, and More](https://docs.oqoqo.ai/reference/supported-surfaces.md): Oqoqo can test any surface an AI agent interacts with: SDKs, REST APIs, CLIs, MCP servers, and agent skills. Learn what each surface tests.
- [Run Scalable Agent Evals Across Models and Treatments](https://docs.oqoqo.ai/use-cases/agent-evals.md): Turn real agent workflows into repeatable evaluations across agents, models, and treatments in production-like environments at scale.
- [Test Agent-Facing Interfaces Before Agents Use Them](https://docs.oqoqo.ai/use-cases/agent-interfaces.md): Measure how AI agents interact with your SDK, API, CLI, or MCP server. Identify friction points and fix them before they affect your users.
- [Build a Benchmark for How Agents Perform on Your Product](https://docs.oqoqo.ai/use-cases/benchmarking.md): Turn your product's user journeys into a repeatable benchmark suite that scores how well AI agents can accomplish real tasks on your product.
- [Block PRs That Break Agent Workflows with CI/CD Gating](https://docs.oqoqo.ai/use-cases/ci-cd.md): Block pull requests that break agent workflows. Integrate Oqoqo into your CI pipeline to catch regressions before they reach production.
- [Detect Agent Workflow Drift from Models and Dependencies](https://docs.oqoqo.ai/use-cases/drift-detection.md): Schedule Oqoqo experiments to run automatically as models, dependencies, and your product evolve — and catch regressions before your users do.
- [Optimize Model Spend by Comparing Agent Performance](https://docs.oqoqo.ai/use-cases/model-spend.md): Find where cheaper models match expensive ones on your workflows. Compare pass rate, cost, and tokens to make confident model-swap decisions.