Agent — The coding harness that does the work in a run (for example Claude Code, Codex, Cursor, Grok Build, Qwen Code, Antigravity, or Hermes), with a model and effort level. Chosen at launch. Enable under Manage → Agents. See Supported agents.Ask Oqo — The in-app assistant for your project. Assign a compatible model under Oqoqo AI features. See Ask Oqo.Assets — Skills, MCP servers, CLIs, and SDKs under Manage → Assets. You attach them to treatments, not to tasks. See Assets.
Clone experiment — Opens a new experiment draft from a finished experiment’s tasks, treatments, and settings. See Launch options.Compare — Side-by-side view of 2 or 3 runs on Traces, Output, and Evals. See Reading results.
Evals — Scoring a finished run against the task rubric. An automated judge marks each requirement pass or fail. Independent of Insights. See Evals and Insights.Experiment — A launched comparison. You select tasks, agents, and treatments, then launch. Oqoqo snapshots what you selected so later catalog edits do not rewrite past results.
Files — Folders you attach to a task. Mounted into the machine before the run starts. Create under Manage → Files. See Machines and files.Friction — A specific point where an agent stalled, retried, or went wrong. Found by Insights. Listed on the experiment Frictions tab.
Insights — Analysis of saved traces and output. Finds frictions and patterns. Independent of Evals. See Evals and Insights.Integrations — Personal connections under Profile → Integrations: GitHub Copilot, OpenAI (ChatGPT Subscription), Anthropic (Claude Subscription), and Nous Portal.
Machine — The environment a run executes in. Pinned on a task. Default is the Standard Dev Machine. Create under Manage → Machines. See Machines and files.Matrix — The default experiment tab. Pass rate, lift, cost, tokens, and duration across tasks, agents, and treatments.Max Concurrency — How many agent sessions can run at once in a launch. Each session is one trial of one task × one agent × one treatment. See Launch options.Metrics — Steps, tool calls, tokens, cost, and duration for a run. They show how hard the agent worked, not only whether it passed.Model providers — The organization catalog under Organization settings → Model providers. A provider can use a shared organization credential or require each person to connect a supported Integration. See Model providers.
Oqoqo AI features — Organization model choices for Ask Oqo, Rubric generation, Run evals, Run insights, and Experiment insights. Compatible organization providers and supported personal Integrations supply the models. See Oqoqo AI features.Output — The run tab for the final assistant message and recorded workspace changes.
Pass rate — The fraction of repeated runs that passed, with a 95% confidence interval.Plan runs — Runs from the free or paid plan for each billing period. The allowance resets each period. Unused plan runs do not roll over.Plugins — Coming soon under Manage → Assets → Plugins. Will bundle skills, agents, hooks, MCP servers, and related capabilities.Project — Your workspace. Holds tasks, files, machines, treatments, assets, agents, and experiments. Managed from the project switcher. See Projects.
Requirement — One pass or fail criterion in a rubric.Re-trigger Evals / Insights — Score or analyze the same recorded work again without re-running the agent. Use when the rubric or analysis models change.Rubric — The pass or fail criteria for a task. Scored by Evals.Run — One agent attempt on a task under a treatment. The unit you pay for. Errored or cancelled runs do not count against your balance. See Billing.Runs — The experiment tab that lists every execution.
Task — The work an agent attempts: instructions, rubric, files, and machine. Create under Manage → Tasks. See Tasks and rubrics.Template — A ready-made experiment draft on Home. See Experiment templates.Top-up runs — Purchased runs that never expire.Traces — The run tab for the full step-by-step trajectory.Transcript — The exportable recorded trajectory for a run. See Traces and output.Treatment — One condition you compare. The raw agent, or the agent plus assets attached to it. Create under Manage → Treatments. See Treatments.Trial — A repeat of the same task × treatment × agent. Each trial is one billable run. Set at launch.
