Skip to content
Kaddo Kaddo

New project

When to use this: you are starting a greenfield project and want to avoid scattered decisions from day one — a lightweight knowledge layer that grows with the code.

Workflow

Terminal window
kaddo init          # state: new, team size, structure
kaddo context       # LLM context pack → .kaddo/context-pack.md
kaddo add agents    # install agent prompt packs
kaddo understand    # guided CLI → LLM handoff plan
# ── in your LLM, use roadmap-agent and architecture-agent to draft
#    architecture/roadmap.md and architecture/current-state.md ──
kaddo create --from roadmap   # turn a roadmap candidate into a Work Item
kaddo owners suggest          # declare code: ownership on the Work Item
kaddo guard                   # detect possible knowledge drift
kaddo explain                 # summarize what Kaddo currently knows

For a brand-new repo you can skip kaddo scan (there is little code to detect yet) and start from the roadmap. Run scan later once the codebase grows.

CLI vs LLM

  • CLI (deterministic): init, context, add agents, understand, create, owners suggest, guard, explain.
  • LLM (interpretation): use the roadmap-agent and architecture-agent in your chat to shape the first roadmap and the intended architecture from the context pack.

Kaddo never calls an LLM — it prepares the context; your LLM does the thinking.

Expected artifacts

.kaddo/config.yml
.kaddo/context-pack.md
.kaddo/understand.md
architecture/roadmap.md
architecture/current-state.md
architecture/work-items/*.md
.kaddo/explain.md

Next steps

Keep creating Work Items from the roadmap, declare ownership as the code lands, and run kaddo guard before commits so knowledge stays connected to code. See the Full workflow for the complete loop.

See it in action: the Task Pilot demo repo, or browse all Examples.