Guide
Team Onboarding
This guide covers the recommended rollout process for deploying FORG across an engineering team — from the first install to org-wide enforcement.
Phase 1: Admin setup (Day 1)
Complete this before any developer installs FORG:
- Create the org — Sign up at forg.pro and create your organization. Choose your plan based on team size.
- Configure SSO (Enterprise) — Set up SSO with your IdP before inviting users. See the SSO setup guide.
- Create teams — Create teams that mirror your org structure (e.g., frontend, backend, infra). Teams are used for scoping rules and budgets.
- Set a baseline budget rule — Create a monthly org-level budget cap at a comfortable ceiling. This can be adjusted later. Start loose.
- Connect Slack — Set up the Slack integration so alerts go to the right channel.
Phase 2: Pilot group (Week 1)
Roll out to 3–5 developers first. This catches integration issues before wide deployment.
- Invite the pilot group via Dashboard → Members → Invite
- Share the install command and have each developer run:
curl -fsSL https://forg.pro/install | sh forg activate <their-license-key> - For Claude Code users, the adapter installs automatically after activation. Have them start a Claude Code session and verify the status line shows the FORG indicator.
- Check Dashboard → Analytics after 2–3 days to verify signals are arriving from pilot users.
Phase 3: Full rollout (Week 2–3)
Once the pilot is stable, roll out to the full team:
- Bulk invite via CSV or SCIM sync. FORG automatically allocates a license seat to each new member.
- Internal comms — Send a short message explaining what FORG tracks (model metadata, tokens, cost — not prompts or completions), how to activate, and who to contact with questions.
- Per-team budgets — Once you have 1–2 weeks of real usage data, set per-team budget rules based on p90 daily spend.
Phase 4: Steady state
After the full rollout, the primary ongoing tasks are:
- Monthly review — Check Analytics → Cost Breakdown. Identify changes in model usage patterns or unusual spikes.
- Rule tuning — Review the Rules → Trigger log. Tighten limits that never trigger; loosen limits that block legitimate work.
- New joiner flow — If using SCIM, new hires are automatically provisioned. Without SCIM, invite new members and include FORG activation in the standard dev machine setup guide.
- Offboarding — When a developer leaves, deactivate their FORG account (via SCIM or manually). Their license seat is released immediately.
Common rollout questions
| Question | Answer |
|---|---|
| Does FORG see what developers type? | No. FORG only sees metadata: model name, token counts, latency, cost. Not prompts or completions. |
| What happens if the FORG agent is not running? | The adapter fails open — AI tools work normally, signals are dropped silently. No interruption to developer workflow. |
| Can developers opt out? | They can stop the agent, but org admins can see gaps in signal data. In enforcement mode, the adapter blocks calls when the agent is unreachable if rules require it. |
| How many license seats do we need? | One seat per active developer who uses an AI coding tool. Seats are per-user, not per-device. |
© 2026 UpgradIQ, Inc.Edit this page on GitHub