Guide for Tech Leads

You're accountable for code quality and team velocity. Fabriq gives you the tools to delegate to AI without losing control — but the setup decisions determine how well it works for your team.

1. Set up the workspace

Before your team creates features, set up the workspace structure:

1. Create the workspace on a shared machine or in a shared directory
2. Register all repositories your team works with — the base repo pool
3. Verify workspace structure — check that .dapei/, docs/, repos/, and features/ exist
4. Establish docs standards — what goes in docs/ (specs, ADRs, design tokens)

→ Workspaces Overview · Adding Repos

2. Configure stage gates as governance

Stage gates are your primary governance tool. Configure them to enforce your team's process:

Gate	What it enforces	Recommended setting
Analysis → Design	Business design doc exists	Hard gate — don't allow coding without a design
Design → Implementation	Technical design doc exists	Hard gate — no code without architecture
Implementation → Validation	testability.md filled in	Soft gate — allow bypass when justified
Validation → Review	Validation run completed	Hard gate — never ship untested code
Review → Close	Feature has been approved	Hard gate — never close unreviewed work

→ Stage Gates

3. Establish review standards

The review stage is where quality assurance happens. Set expectations:

• Validation badges must be green before merging — lint, types, tests all passing
• Known failures require a comment — if you approve with failing tests, the reason is recorded
• Gate bypasses are logged — every confirmed=true override appears in the audit trail
• Feature descriptions should be peer-reviewed before the agent starts working

4. Use the audit trail for compliance

The audit trail is your record of "who approved what, and when." Use it for:

• Post-ship review — after a release, audit the features that shipped
• Incident investigation — when something breaks, check the audit trail for the approving transaction
• Process metrics — track time between stages, identify bottlenecks

→ Audit Trail

5. Onboard a new engineer

1. Send them the Getting Started section (15 minutes)
2. Have them run through Your First 15 Minutes on a test workspace
3. Pair on their first real feature — review the confirmation prompts together
4. After their first close, audit the trail together — show them what's recorded

6. Monitor and iterate

Fabriq's audit trail gives you visibility into your team's AI-assisted workflow. Watch for:

• Frequent gate bypasses — if engineers consistently bypass a gate, the process may not match reality
• Long review cycles — agents producing work that needs major rework → invest in better feature descriptions
• Underutilized validation — if features skip to review without validation, reinforce the gate
• Stale context docs — if agents make mistakes about architecture, update runtime-context.md

Next steps

• Guide for Engineers — understand your team's daily workflow
• Writing Great Feature Descriptions — teach the team to write better briefs
• Stage Gates — detailed gate configuration
• Audit Trail — compliance and monitoring