Orizu CLI

Overview

Execute end-to-end Orizu workflows from the terminal using the orizu CLI. Prefer explicit flags in automation and CI; use interactive prompts only in TTY sessions.

Prerequisites

• Ensure Node.js 20+ is installed.
• Ensure Orizu API is running (http://localhost:3000 by default) or set ORIZU_BASE_URL.
• Build CLI from source when needed:

  bun install
  bun x tsc -p packages/cli/tsconfig.json
  node packages/cli/dist/index.js --help
  

• Use orizu directly when globally installed; otherwise run node packages/cli/dist/index.js ....

Core Workflow

1. Authenticate:

   orizu login
   orizu whoami
   

2. Set up workspace:

   orizu teams create --name "Ops Eval"
   orizu projects create --name "Support QA" --team ops-eval
   

3. Upload dataset:

   orizu datasets upload --project ops-eval/support-qa --file ./datasets/support.jsonl --name "Support Batch 1"
   

4. Optionally append new rows later:

   orizu datasets append --dataset <datasetId> --file ./datasets/support-additional.jsonl
   

5. Optionally delete incorrect rows:

   orizu datasets delete-rows --dataset <datasetId> --row-ids <rowId1,rowId2>
   

6. Create or update app from file (dataset is required):

   orizu apps create \
     --project ops-eval/support-qa \
     --name "Support Labeler" \
     --dataset <datasetId> \
     --file ./apps/SupportLabeler.tsx \
     --input-schema ./schemas/support-input.json \
     --output-schema ./schemas/support-output.json
   

7. Optionally link a different dataset to an existing app version:

   orizu apps link-dataset --app <appId> --dataset <datasetId>
   

8. Run task lifecycle (task create requires assignees and creates assignments immediately):

   orizu tasks create --project ops-eval/support-qa --dataset <datasetId> --app <appId> --title "Support QA Round 1" --assignees <userId1,userId2>
   orizu tasks status --task <taskId>
   orizu tasks export --task <taskId> --format csv --out ./support-round1.csv
   

Interactive vs Non-Interactive

• Use interactive fallback only when running in a TTY and flags are omitted.
• Provide explicit identifiers in scripts/CI:
  • --team, --project, --app, --task, --dataset, --assignees.
• For tasks assign, pass user IDs (comma-separated), not emails.
• Export defaults:
  • --format defaults to jsonl.
  • output defaults to <taskId>.<format>.
• Dataset row mutation:
  • datasets append accepts --file in .csv, .json, or .jsonl.
  • datasets delete-rows requires at least one of --row-ids or --row-indices.

Auth and Error Handling

• On auth failure, run orizu login and confirm with orizu whoami.
• Login requires localhost callback availability on 127.0.0.1:43123.
• Credentials are stored in ~/.config/orizu/credentials.json.
• In non-interactive contexts, avoid prompt-dependent commands; pass required flags explicitly.

Command Coverage

• Auth: login, logout, whoami
• Teams: list/create, members list/add/remove/role
• Projects: list/create
• Apps: list/create/update/link-dataset
• Datasets: upload/download/append/delete-rows (.csv, .json, .jsonl)
• Tasks: list/create/assign/status/export

References

• Read references/cli-reference.md for complete command examples and end-to-end flows.
• Read docs/cli.md for the authoritative project-local CLI guide when working in this repository.