Skip to content

Example workflows

Take a ticket through delivery

With tracker and SCM connections configured:

  1. Draft work from an existing issue: ai-dlc work new improve-setup --from-issue 42
  2. Edit .ai-dlc/work/improve-setup.toml: set scope and acceptance criteria, decide whether a specification is required and link it when needed. Then record the review: ai-dlc work review improve-setup. It refuses leftover TODOs and records who reviewed it and when.
  3. Publish and start it: 
    ai-dlc work publish improve-setup
ai-dlc work start improve-setup
  4. Implement, commit, and update the affected documentation. Then run ai-dlc project check --required. If this work requires OpenSpec, finalize it with ai-dlc work archive before review.
  5. Push the bound branch, then open its PR: ai-dlc work pr improve-setup
  6. Push the work-record link commit that work pr created, then review and merge.
  7. From a checkout at the merge commit, after CI succeeds, run ai-dlc work finish improve-setup.

Finish validates the required specification, PR merge and exact merged-revision CI evidence, plus configured deployment gates. work status is a local inspection, not a fresh tracker read.

Keep project documents useful

ai-dlc docs init                  # Preview missing navigation.
ai-dlc docs check                 # Inspect ownership, coverage, and links.
ai-dlc docs search "authentication"
ai-dlc docs review --base origin/main

Use reviewed dispositions and ai-dlc docs gate to keep documentation current with a change. Project-document access covers repository files; private vault access uses separate knowledge tools. The documentation guide covers review, MCP search/read, and optional Obsidian portals and mounts.

Start an FDE engagement

ai-dlc fde scaffold customer-delivery --title "Customer delivery" --dry-run
ai-dlc fde scaffold customer-delivery --title "Customer delivery"
ai-dlc fde check customer-delivery

This creates a charter and Discover → Frame → Design → Build → Deploy → Enable → Expand landing pages under docs/fde_engagements/. Stage checks require earlier exit criteria and recorded evidence before downstream activation. See the FDE guide for configuration and stage metadata. Confluence publication and synchronization remain unimplemented.