Skip to main content

Troubleshooting

This page collects the most common issues across local development, sign-in, integrations, cloud agents, and corporate networks. Search the docs first — press ⌘K (or Ctrl K) — then work through the relevant section below.

Search the docs

Press ⌘K (or Ctrl K) and type what you're trying to do. If search doesn't open, refresh the page and try again.

Local development ports

Each app uses a fixed port so nothing collides. If a dev server fails to start with a port error, stop the process using that port or pick the correct app:

PortAppStart command
3000Marketing site (web)npm run dev:web
3001Dashboardnpm run dev:dashboard
3002Docsnpm run dev:docs
3003Authenticatornpm run dev:auth
3004Cloud agents UInpm run dev:agent
3005Marketplace registrynpm run dev:registry
3006Consolenpm run dev:console
3007Platform APInpm run dev:platform-api
3008Agents APInpm run dev:agents-api
3009Projectsnpm run dev:projects
Find what's using a port

On macOS: lsof -i :3008 then kill <PID>. On Linux, use ss -tlnp | grep 3008.

Common local setup mistakes

SymptomLikely causeFix
fetch failed on sign-inAuthenticator not runningStart port 3003; check NEXT_PUBLIC_AUTH_URL
Agent runs 404Agents API not runningStart port 3008
Dashboard policy emptyPlatform API not runningStart port 3007
Docs links 404 in devWrong portDocs are on 3002, not 3000

Sign-in and authentication

  1. Confirm authenticator is running

    The authenticator app must be reachable at the URL in your app's NEXT_PUBLIC_AUTH_URL (port 3003 locally, auth.4rged.ai in production).

  2. Verify Supabase env vars

    Each app needs NEXT_PUBLIC_SUPABASE_URL and NEXT_PUBLIC_SUPABASE_ANON_KEY pointing at the same Supabase project.

  3. Check org membership

    After sign-in, if you land on a no-access page, ask an admin to invite you from Members.

  4. Desktop handoff

    If the browser doesn't return to the IDE, see Deeplinks.

ErrorCauseFix
Can't sign in / fetch failedAuth app down or wrong URLStart authenticator; verify env
No access after sign-inNot in an organizationRequest invite from org admin
Console shows but expected dashboardPlatform role routingExpected for platform operators
Desktop bridge "Missing handoff"Empty token fragmentRestart sign-in from IDE

Integrations

GitHub

  • Shows connected but repos don't sync locally — full repository sync requires the Supabase service role key (SUPABASE_SERVICE_ROLE_KEY) in your local agents-api environment. Without it, the GitHub app connection succeeds but repository metadata sync is limited.
  • Webhook deliveries failing — confirm GITHUB_WEBHOOK_SECRET matches the GitHub app config and the agents API is reachable from GitHub (use ngrok locally).
  • PR creation denied — verify the GitHub app has write access to the repository and branch protection allows the bot account.

Slack and Linear

Partial integrations

Slack or Linear "not connected yet" — only GitHub is fully wired in the dashboard today. Other integrations appear in the UI and automation trigger picker but are not connectable yet. Use GitHub or schedule triggers for production automations.

IntegrationStatusWorkaround
GitHubFully connectable
SlackUI onlyMonitor runs in Agents workspace; use GitHub notifications
LinearUI onlyPaste issue URLs into cloud agent prompts manually

See Slack and Linear for planned capabilities.

Cloud agents

SymptomCauseFix
Run denied with 402Budget or contract limitCheck Usage; ask operator to adjust capacity
Stuck in queuedWorker backlog or env issueCheck operations summary; see Sentinel
Setup command failsWrong install script or blocked registryFix setup commands; allowlist npm/PyPI domains
npm ci network errorEgress blockedAdd registry to allowlist
No PR openedMissing GitHub write permsReinstall app with repo write access

Start a debug run

curl -X POST "http://localhost:3008/api/agent-runs" \
  -H "Authorization: Bearer <session-access-token>" \
  -H "Content-Type: application/json" \
  -d '{
    "organizationId": "org_local",
    "repositoryId": "1",
    "prompt": "Print node version and exit",
    "agentKind": "cloud"
  }'

Cloud agents and network

If you're behind a corporate firewall or proxy, outbound traffic to 4RGE services may be blocked. Allowlist these domains (HTTPS unless noted):

DomainPurpose
*.4rged.aiAll 4RGE product and API surfaces
*.supabase.coAuthentication and platform data
registry.npmjs.orgnpm installs in cloud sandboxes
pypi.org, files.pythonhosted.orgPython installs
github.com, api.github.comGitHub integration

See Security & network for the full allowlist and operator checklist. Ask your network admin to test from the same VLAN as developer machines.

Console and provisioning

SymptomCauseFix
403 on provisioning routesMissing platform roleConfirm platform operator access
Contract update rejectedInvalid kind or missing reasonSee Contracts API shape
Capacity change not visibleOrg admin cacheHard refresh Usage page

Docs site validation (contributors)

If you edit documentation in apps/docs/content, run npm run validate:content from apps/docs before opening a PR. The script checks _meta.json navigation manifests, verifies every listed page exists, and validates MDX frontmatter (title, optional description) with Zod. Missing files or empty titles fail CI locally — fix manifest paths before pushing. The docs app serves on port 3002 (npm run dev:docs); production docs live at docs.4rged.ai.

Still stuck?

See Help & support for who to contact — your org admin for access and policy questions, or support@4rge.ai for everything else.

When contacting support, include:

  • Organization ID and affected repository
  • Run ID or automation ID (if applicable)
  • Timestamp and error message
  • Whether you're on a corporate network or VPN