Forge is preparing the requested surface and verifying the live route.
Forge is preparing the requested surface and verifying the live route.
Forge Agents
The website explains the agent model. Real chat, coding, workbench, living codebase, provider access, and account-powered execution happen inside the Forge apps.
Forge agents need runtime ownership, credential scoping, policy checks, mission continuity, and receipts. That makes the desktop, mobile, and signed-in workspace the right place to operate them.
Forge agents operate inside the app/workspace that owns provider credentials, local runtime access, permissions, receipts, and context.
Skills, tools, memory, and routing are enabled through policy and capability ledgers rather than public marketing-page controls.
Multi-agent work uses durable mission state, workbench evidence, and owner-visible next actions instead of hidden public-site automation.
Agents can work against the living codebase map from the app workbench where the repository, graph, and execution receipts are current.
Ask an agent to inspect a repo, propose a change, run tests, and preserve proof from the coding pane.
Have agents gather evidence, cite sources, and route follow-up work into the same mission object.
Coordinate work across local devices, providers, and runtime owners with visible blockers and receipts.