nWave-ai Command Catalog
| Need | Command |
|---|---|
| Start any delivery request | printf '%s' "$REQUEST" | des po --repo-root ROOT |
| Continue, or resume after an interruption | des state --repo-root ROOT, then have the LLM assess the advisory NEXT and current evidence before choosing a supported operation |
| Read the state as a person | des project --repo-root ROOT |
| Construct reviewer or examiner input after native verify | des prepare-role --repo-root ROOT --role reviewer|examiner --candidate SHA |
| Record an LLM-hosted role result | des record-role-result --repo-root ROOT --role reviewer|examiner --candidate SHA --provider PROVIDER --model MODEL --session-id SESSION --input - |
| Explicitly invoke one configured reviewer or examiner | des invoke-role --repo-root ROOT --role reviewer|examiner --candidate SHA --provider PROVIDER --input INPUT |
| Select an explicit Codex or Claude model for DES | Use nw-rigor, then run its matching nwave-ai model set --provider PROVIDER --model MODEL [--role ROLE] [--project] command |
Every step returns Success, Refusal, Retry, or Indeterminate, and may name a successor as advisory data in NEXT. For features and bugfixes, the LLM evaluates and revises S/M/L and chooses applicable waves or upstream correction from evidence; DES neither maps size to a route nor executes NEXT. No command composes the steps into a sequence.