gpt2giga Harness

Proxy: checking

Workspace Harness Include archived

Native sessions

Show all workspaces

Native history not synced

Harnesses

What do you want to work on?

Ask a question, plan a task, or describe what you would like to build.

Harness Invocation Model

API mode

/v2 /v1

Current route: /v2/chat/completions

Mode Workspace policy Permission profile Capability Workspace

dry run stream

Recommended: pending

Type a prompt or attach context to refresh the recommendation.

Presets

Presets: loading

Prompt

Try an example to get started

Run details

No session selected

No run selected.

This run is not part of an agent team.

No events yet.

{}

{}

No command yet.

No diff captured.

No PR artifact.

No provenance selected.

No editor action yet.

No attachments selected.

Memory: loading

Project memory

Run tools: none Open Tools

Select a run to inspect its tool profile summary.

MCP inventory, schemas, health, and compatibility live in Tools.

Scorecard: loading Open Evaluate

No eval run selected.

Native: idle

Native CLIs stay active between turns. Terminal output streams automatically, polling remains available as a fallback, and Stop process ends the session.

No native session selected.

Technical terminal output

Terminal output will appear here.

No storage selected.

gpt2giga Harness

Native sessions

Harnesses

What do you want to work on?

Presets

Arena

Responses

Agent Studio

Select an agent

Workflow Catalog

Select a workflow

Steps

DAG preview

Scheduled Automation

New schedule

Approval Center

Eval Lab

Protocol conformance

Harness quality

Runs and trends

Tools

Runs Center

Select a run