kaddo init
kaddo initCreates the knowledge/ knowledge tree and .kaddo/config.yml.
knowledge/ knowledge.md ← current state of the product roadmap.md ← intentions and priorities work-items/ ← one file per work item.kaddo/ config.yml ← project configProject context
kaddo init asks three questions and records the answers in .kaddo/config.yml.
Other commands read this file to adapt their guidance — Kaddo becomes state-aware.
version: 1project: name: "my-app" state: pre-ai # new | pre-ai | legacy structure: monorepo # monorepo | multirepo language: en # en | es — language of the project KNOWLEDGE domains: []team: size: indie # indie | small | medium | enterpriseProject state — drives next-step guidance in scan and create:
| State | Meaning |
|---|---|
new | Greenfield project; start defining knowledge from scratch. |
pre-ai | Existing system without an AI/knowledge baseline yet. |
legacy | Older system where risks and unknowns matter most. |
Team size — indie, small, medium, enterprise.
Repository structure — monorepo or multirepo.
Project language — en or es. This is the language of the project knowledge (templates,
agent outputs, context pack, roadmap, Work Items, ADRs, capabilities…), not the CLI. The CLI —
its commands, flags, prompts and messages — is always English. See
Project knowledge language.
Old config files keep working: missing fields fall back to safe defaults
(state: pre-ai, structure: monorepo, team.size: indie, language: en). Invalid enum values
produce a clear validation error listing the valid options.