Setup Web UI (Preview)
Run the local setup UI:
intelligencex setup webThis starts a local web server and opens the wizard in your browser (http://127.0.0.1 only).
Screenshots
- Configure step: Screenshot
- Verify step: Screenshot
Quick flow
1) Start the web UI
2) Choose onboarding path (new setup / fix auth / cleanup / maintenance)
3) Run auto-detect preflight
4) Authenticate with GitHub (device flow or app install)
5) Select repos
6) Plan + Applyflowchart LR
classDef step fill:#BAE6FD,stroke:#0369A1,color:#082F49,stroke-width:2px;
classDef auth fill:#FDE68A,stroke:#B45309,color:#451A03,stroke-width:2px;
classDef apply fill:#A7F3D0,stroke:#047857,color:#052E2B,stroke-width:2px;
A["Path"] --> B["Auto-Detect"]
B --> C["GitHub Auth"]
C --> D["Repos"]
D --> E["Configure/Auth"]
E --> F["Plan"]
F --> G["Apply + Verify"]
class A,B,D,E,F step;
class C auth;
class G apply;Path requirements (GitHub/repo/AI auth) and Bot contract checks are defined in Web Onboarding Flow .
Operations available:
- Setup / update workflow + config
- Update OpenAI secret only (requires auth bundle)
- Cleanup (remove workflow/config)
- Maintenance (inspect and choose operation)
- Optional GitHub App manifest flow (create app + installation token)
- Load existing config from a repo (manage existing setup)
- Load workflow preview for the managed workflow
- Save/load config presets in the browser
- Export/import presets as JSON files
- Import prompts before overwriting existing presets
Advanced options:
- Provider toggle (openai | copilot)
- Static analysis controls when generating preset config (
analysisEnabled,analysisGateEnabled,analysisRunStrict, packs, export path) - OpenAI account routing supports primary-only setup (rotation/failover can be configured without
account ids) - Auth bundle input for secret updates (INTELLIGENCEX AUTH B64)
For YAML vs JSON ownership and precedence, see Workflow vs JSON .
GitHub App flow (optional)
If you want to avoid personal access tokens, you can use the GitHub App manifest flow:
- Enter App name + App owner (org login).
- Click “Create App (manifest)”. A browser window opens to create the app.
- Install the app in the org/user and return to the wizard.
- Click “List installations”, select the installation, then click “Use installation token”.
- The GitHub token field is populated with the installation token; proceed to load repos.
Current limitations
- OpenAI login runs locally in the web UI ("Sign in with ChatGPT") and returns an auth bundle (
authB64) you can upload asINTELLIGENCEX_AUTH_B64. - Update-secret in the web UI requires an auth bundle (either from the login button or pasted via
authB64/authB64Path). - The UI supports multi-repo setup (plan/apply), repo inspection, and setup recommendations.
- GitHub App installation tokens can only list repos the app is installed on.
- Buttons are disabled until required inputs are provided (token, repo selection, auth bundle).
- Inline hints describe what is missing before plan/apply can run.
- Status badges show auth, repo selection, and auth bundle readiness.
Tips
- Use the "Load workflow preview" button before applying changes.
- If you want zero secret handling in the UI, enable "Skip OpenAI secret" and paste secrets manually in GitHub.
- Start with auto-detect to get a recommended path before selecting repositories.
- If you automate setup with Bot tools, verify
contractVersion+contractFingerprintmatch before apply.
Security notes
- The server listens on 127.0.0.1 only over HTTP (no HTTPS binding).
- GitHub tokens are sent to the local CLI server and never leave your machine.
- Use this on trusted machines only.