Documentation
Learn the runtime before you wire a UI on top of it.
This docs area is organized for a new visitor: how to start, what mental model to adopt, and which workflows matter once agents are doing machine-side work.
Start Here
Documentation overview
The full documentation map for Gooselake: setup, mental model, runtime services, client APIs, operations, deployment, and reference material.
Open guideSetup
Install Gooselake, run the runtime locally, and verify the first control-plane loop.
Open guideInstall guide
Install Gooselake from a release artifact or from source, including service files, provider CLIs, upgrades, and bundle layout.
Open guideCLI and command runner
Understand the current gg-runtime-server command-line surface, Make targets, scripts, and what is still HTTP-only.
Open guideMental Model
Core concepts
Understand the runtime boundary, event model, provider abstraction, host-owned execution, teams, processes, and worktrees.
Open guideRuntime lifecycle
Follow sessions, turns, approvals, provider dispatch, terminal states, and close/resume behavior through the runtime.
Open guideEvents and recovery
Understand replayable event streams, cursors, criticality, SSE handoff, startup recovery, and diagnostics.
Open guideArchitecture
Understand the runtime control-plane boundary, core crates, provider adapters, sidecars, persistence, events, and recovery model.
Open guideRuntime Services
Provider guide
Configure and diagnose Codex, Claude, and ACP behind the shared Gooselake runtime provider contract.
Open guideTeams and comms
Use Gooselake teams, members, durable messages, delivery policies, retries, cancellation, and spawn operations.
Open guideProcesses
Run host commands through the runtime, stream sampled output, read authoritative logs, and understand process ownership.
Open guideWorktrees
Create, claim, release, repair, and clean up managed Git worktrees for runtime sessions and spawned team members.
Open guideMCP and sidecars
Understand how Gooselake uses sidecar processes for Claude bridge behavior and the bundled MCP tool server.
Open guideClient Builders
Usage model
Learn how clients, operators, scripts, and provider sessions interact with the runtime during daily use.
Open guideClient design guide
Build thin frontends, local consoles, scripts, and future CLIs on top of Gooselake without duplicating runtime state.
Open guideAPI guide
Use the Gooselake HTTP and SSE API for sessions, turns, events, approvals, providers, processes, worktrees, teams, diagnostics, and MCP gateway calls.
Open guideOperators
Configuration reference
Configure server binding, auth, data paths, providers, events, processes, worktrees, sidecars, and environment overrides.
Open guideOperations runbook
Run health checks, inspect recovery, debug providers, manage processes and worktrees, and recover runtime state after restarts.
Open guideDeployment guide
Deploy Gooselake as a host-side service with staged releases, systemd, reverse proxies, provider auth, and upgrade/rollback paths.
Open guideSecurity model
Understand bearer auth, local trust boundaries, provider credentials, process execution, MCP caller identity, and deployment precautions.
Open guideTroubleshooting
Debug common Gooselake failures around auth, providers, sessions, streams, teams, processes, worktrees, MCP, and docs sync.
Open guideOperator workflows
The workflows that matter once Gooselake is running real agent work on a real machine.
Open guideReference
Endpoint catalog
Browse the implemented runtime HTTP and SSE route surface, including public routes, bearer-protected routes, and provider-specific endpoints.
Open guideRepo guide
Use the repository layout and existing docs to move from orientation into implementation and operations.
Open guideAPI doc sync workflow
Keep the generated OpenAPI artifact and human API docs synchronized when runtime route behavior changes.
Open guideChangelog
Release notes for Gooselake.
Open guide