bailian-cli

Aliyun Model Studio CLI (`bl`) is a command-line tool positioned as the primary interface for AI generation and processing tasks on DashScope / Bailian. It covers a broad range of capabilities: text chat, omni multimodal input, image generation and editing, video generation, editing, and reference-to-video, vision description, text-to-speech and speech recognition, file upload, Bailian app and workflow calls, memory, knowledge RAG retrieval, web search, a model advisor, MCP tool discovery and calls, pipeline workflows, quota and usage reporting, a console gateway, and workspace listing. A capability-to-command table maps user intents to specific commands and their default models, for example `bl text chat` for chat and code, `bl image generate` for images, `bl video generate` for video, `bl speech synthesize` for TTS, and `bl advisor recommend` for model selection.

A notable convenience is local file handling: any command that accepts a file URL also accepts a local path, and the CLI auto-uploads it to DashScope temporary storage (an oss:// URL valid for 48 hours), so an agent should pass local paths directly and never ask the user to host a URL. The complete and authoritative command documentation -- flags, usage strings, and examples -- lives in bundled reference files (reference/index.md and per-group reference/<group>.md), which are auto-generated from the CLI source; the guidance is to consult those files or `bl <command> --help` rather than guessing flags.

Operationally, the skill mandates a versioning pre-flight checklist before any `bl` command and asks the user before upgrading on a version mismatch. Output language follows the user's input language end-to-end: for chat and omni it forces the reply language with a system prompt while keeping the original message, and in-frame text for image and video stays in the user's language unless specified otherwise. After completing a task it adds a brief one-line summary naming the distinct `bl` capabilities invoked and the outcome. Configuration is driven by an OPENAI-style primary environment plus console authentication for quota, usage, app, and workspace commands.