
Aigon - Spec-Driven Development for AI Agents

Usage: aigon <command> [arguments]
       aigon --version

Setup:
  install-agent <agents...>         Install agent configs ({{SUPPORTED_AGENT_INSTALL_IDS}})
  install-seed <name>               Clone a demo seed repo (brewboard, trailhead) into ~/src/<name>, apply Aigon, and register it with the dashboard
  apply                             Initialize repo on first run, then keep Aigon files up to date (re-vendor docs, refresh agent configs)
  remove [--dry-run] [--force] [--purge]  Remove Aigon-managed files from this repo; deregisters from dashboard
  hooks [list]                      List defined hooks (from docs/aigon-hooks.md)
  config <init|show|models>         Manage config and view model configuration
  profile [show|set|detect]         Manage project profile (web, api, ios, etc.)
  doctor [--register]               Check port assignments and model configuration
  proxy install                     One-time: install Caddy proxy as system daemon on port 80
  proxy <start|stop|status|uninstall> Manage the Caddy proxy daemon

Dev Server (web/api profiles):
  dev-server start [--port N] [--open]  Start dev server, add Caddy route, wait for healthy
  dev-server start --register-only  Register port mapping only (don't start process)
  dev-server stop [serverId]        Stop process and remove Caddy route
  dev-server open                   Open dev server URL in default browser
  dev-server list                   Show all active dev servers across all apps
  dev-server logs [-f] [-n N]       Show dev server output (default: last 50 lines, -f to follow)
  dev-server gc                     Remove entries for dead processes
  dev-server url                    Print URL for current context (for scripting)

Backup & Sync (aigon-vault):
  backup configure [<git-url>]      Point one private vault repo at all your aigon data (interactive, can create on GitHub)
  backup push                       Back up every project's state + global settings to the vault
  backup pull                       Restore settings and project state from the vault on a fresh machine
  backup status                     Show remote, last push/pull, schedule, and project count
  backup schedule <cadence>         daily | hourly | weekly | off  (default: daily, runs in the background)

  sync …                            Deprecated alias for `backup` — use `aigon backup` instead

Modes:
  🚗 Drive      One agent, you guide each stage
  🚛 Fleet      Multiple agents in parallel, you observe and guide
  ✈️  Autopilot  One agent runs autonomously end-to-end
  🐝 Swarm      Multiple agents run autonomously in parallel

Conductor:
  feature-autonomous-start <ID> <agents...>                [Pro]
                                    Start AutoConductor (tmux auto session) and return immediately
  feature-autonomous-start status <ID>                     [Pro]
                                    Show AutoConductor tmux liveness + workflow state
  server start [--persistent]       Start the AIGON server (dashboard UI + supervisor)
  server add [path]                 Register a repo with the server

Feature Open:
  feature-open <ID> [agent] [--terminal=<type>]
                                    Open feature worktree in terminal and start agent (warp, code, cursor, terminal, tmux)
  feature-open <ID> --all           Open all Fleet worktrees side-by-side
  feature-open <ID> <ID>... [--agent=<code>]
                                    Open multiple features side-by-side
  sessions-close <ID>               Kill all agent sessions for a feature/research ID and close Warp/tmux sessions
  session-list                      Print all live tmux sessions Aigon manages (entity + repo categories)

Feature Commands (Drive, Fleet, Autonomous, Swarm):
  feature-create <name>             Create feature spec in inbox
  feature-now <name>                Fast-track: inbox match → prioritise + setup + implement; or create new + implement
  feature-prioritise <name>         Move feature from inbox to backlog (assigns ID)
  feature-start <ID> [agents...]    Start Drive (branch) or Fleet (worktrees)
  feature-do <ID> [--agent=<id>] [--iterate]  Do the work on a feature; launches agent from shell (--iterate runs Autopilot retry loop)
  feature-spec <ID> [--json]        Resolve the canonical visible spec path for a feature
  feature-status <ID> [--json]      Show deep status: session, progress, cost, spec criteria
  feature-list [--active] [--all] [--json]
                                    Query feature records without using the board renderer
  set list [--all] [--json]         List feature sets (spec `set:` frontmatter tags) with stage counts
  set show <slug> [--json]          Show set members in dependency order + intra-set edges
  feature-validate <ID>             Evaluate acceptance criteria (smart validation)
  feature-eval <ID> [--force]       Create evaluation (code review or comparison)
  feature-code-review <ID>          Code review with fixes by a different agent
  feature-code-revise [ID]          Implementer-side review follow-up (infers ID from worktree branch)
  feature-push [ID] [agent]         Push feature branch to origin for PR review
  feature-close <ID> [agent]        Merge and complete feature
  repair <feature|research> <ID> [--dry-run]
                                    Repair safe drift in a feature or research item
  feature-cleanup <ID>              Clean up Fleet worktrees and branches (worktrees + branches only)
  feature-reset <ID>                Full reset back to backlog: sessions + worktrees + branches + state + spec move
                                    (use this to start a feature over with a different agent)
  feature-autonomous-start <ID> <agents...> [--eval-agent=<id>] [--stop-after=implement|eval|close] [--workflow=<slug>]   [Pro]
                                    Start autonomous execution in dedicated auto tmux session

Workflow Definitions (autonomous orchestration templates):
  workflow list [--json]            List workflows (built-in + global + project)
  workflow show <slug> [--json]     Show workflow stages and resolved autonomous inputs
  workflow create <slug> [opts]     Create a project (default) or --global workflow
                                    opts: --file=<path> | --json=<inline> | --from=<slug> |
                                          --implement=<a,b> [--review=<a>] [--revision=<a>]
                                          [--eval=<a>] [--no-close] [--label=<s>] [--description=<s>]
  workflow delete <slug> [--global] Delete a user-created workflow (built-ins are read-only)

Research Commands:
  research-create <name>            Create research topic in inbox
  research-prioritise <name>        Move research from inbox to backlog (assigns ID)
  research-start <ID> [agents...]   Start Drive (no agents) or Fleet (with agents) research
  research-open <ID>                Open all Fleet agents side-by-side for parallel research
  research-do <ID>                  Do research (agent writes findings)
  research-eval <ID>                Evaluate Fleet findings and recommend features
  research-close <ID>               Complete research (in-evaluation → done)
  research-autopilot <ID> [agents...] Fleet autopilot: setup + spawn + monitor + evaluate    [Pro]

Feedback:
  feedback-create <title>           Create feedback doc in inbox (assigns next ID)
  feedback-list [filters...]        List feedback items by status/type/severity/tag
  feedback-triage <ID> [options]    Preview and apply triage updates (requires --apply --yes)

Deploy:
  deploy                            Run the configured deploy command (production)
  deploy --preview                  Run the configured preview command (staging)
  insights [--coach] [--refresh]    AADE Insights    [Pro]

Visualization:
  board                             Show Kanban board view of features and research
  commits [--feature <id>] [--agent <id>] [--period <Nd|Nw|Nm>] [--limit <N>]
                                    Show commit activity with feature/agent attribution
  board --list                      Show detailed list view (features and research)
  board --features                  Show only features
  board --research                  Show only research
  board --active                    Show only in-progress items
  board --all                       Include done items

Context-Aware:
  next                              (agent-only) Detect current context and suggest next workflow action

Token Window:
  token-window [--message=<text>] [--agents=<id1,id2>] [--dry-run]
                                    Send a kickoff nudge to active agent sessions to align provider usage windows

Scheduled kickoffs (server poller; requires aigon server):
  schedule add feature_autonomous <id> --run-at=<iso8601> <agents...> [--stop-after=] [--repo=]
  schedule add research_start <id> --run-at=<iso8601> [agents...] [--repo=]
  schedule list [--all] [--repo=]   schedule cancel <jobId> [--repo=]

Examples:
  aigon apply                          # Initialize repo (first run) or refresh templates
  aigon install-agent cc gg            # Install Claude and Gemini configs

  # Feature workflow
  aigon feature-create "dark-mode"     # Create new feature spec
  aigon feature-now dark-mode          # Fast-track: inbox match or create new + implement
  aigon feature-prioritise dark-mode   # Assign ID, move to backlog
  aigon feature-start 55               # Drive mode (creates branch)
  aigon feature-start 55 cc gg cx cu         # Fleet mode (creates worktrees)
  aigon feature-autonomous-start 42 cc            # Solo autonomous flow (implement -> close)
  aigon feature-autonomous-start 42 cc gg --eval-agent=gg --stop-after=eval
                                                   # Fleet autonomous flow (implement -> eval)
  aigon feature-autonomous-start status 42         # Check conductor status
  aigon server start                   # Start AIGON server (dashboard UI + API)
  aigon feature-open 55 cc             # Open feature worktree and start Claude
  aigon feature-open 55 --all          # Open all Fleet agents side-by-side
  aigon feature-open 100 101 102       # Open features side-by-side (parallel)
  aigon sessions-close 55              # Kill all Fleet agents + tmux sessions + close Warp tab
  aigon feature-do 55             # Launch default agent (cc) from plain shell
  aigon feature-do 55 --agent=cx  # Launch Codex from plain shell
  aigon feature-do 55 --iterate               # Run Autopilot loop
  aigon feature-do 55 --iterate --max-iterations=8 --agent=cx  # Autopilot with options
  aigon feature-validate 55            # Evaluate acceptance criteria (smart validation)
  aigon feature-validate 55 --dry-run  # Show what would be checked without running
  aigon feature-eval 55                # Evaluate implementations
  aigon feature-close 55 cc             # Merge Claude's Fleet implementation
  aigon feature-cleanup 55 --push      # Clean up losing Fleet branches
  aigon feature-reset 55               # Full reset back to backlog (start over with a different agent)
  aigon commits --period 30d           # Commit activity in the last 30 days
  aigon commits --feature 145 --agent cx # Filter by feature and agent

  # Port health check
  aigon doctor                        # Show port assignments, flag conflicts
  aigon doctor --register             # Register current project in global registry

  # Dev proxy (web/api projects) — uses .localhost domains, zero DNS config
  aigon proxy install                 # One-time: system daemon on port 80, auto-starts on boot
  aigon proxy status                  # Check proxy status
  aigon dev-server start              # Register dev server → http://cc-119.myapp.localhost
  aigon dev-server start --open       # Start and open browser automatically
  aigon dev-server open               # Open dev server URL in browser
  aigon dev-server list               # Show all running dev servers
  aigon dev-server stop               # Deregister current dev server
  BASE_URL=$(aigon dev-server url) npx playwright test  # E2E tests

  # Backup & sync (new machine setup)
  aigon backup configure                # Interactive — offers to create aigon-vault on GitHub
  aigon backup push                     # Back up every project + global settings
  aigon backup pull                     # Restore everything on another machine
  aigon backup schedule daily           # Default: daily auto-push runs in the background

  # Research workflow
  aigon research-create "api-design"   # Create new research topic
  aigon research-prioritise api-design # Assign ID, move to backlog
  aigon research-start 05              # Drive mode (one agent)
  aigon research-start 05 cc gg        # Fleet mode (multiple agents)
  aigon research-open 05               # Open all Fleet agents side-by-side
  aigon research-do 05                 # Agent conducts research
  aigon agent-status research-complete 05 cc   # Signal cc's findings are complete
  aigon research-autopilot 08 cc gg cx cu    # Fleet autopilot: spawn + monitor + evaluate
  aigon research-eval 05               # Evaluate findings (after Fleet)
  aigon research-close 05              # Complete research (in-evaluation → done)

  # Deploy
  aigon deploy                         # Deploy to production (uses commands.deploy or scripts.deploy)
  aigon deploy --preview               # Deploy to staging/preview

  # Feedback workflow
  aigon feedback-create "Save fails"   # Create feedback in inbox
  aigon feedback-list --inbox          # List inbox feedback
  aigon feedback-list --all --tag auth # Filter by tag
  aigon feedback-triage 14             # Preview triage suggestions
  aigon feedback-triage 14 --type bug --severity high --tags auth,regression --apply --yes

Agents:
{{SUPPORTED_AGENT_HELP_LINES}}


Pro features are currently in development. Commands marked [Pro] will be enabled when Pro launches.

Documentation: https://www.aigon.build/docs
