What problem does it solve?
Agents often need to call DeepChat capabilities—model inference, image/video/speech generation, transcription, OCR, and configuration queries—that are not exposed as dedicated tools. This Skill provides a governed command-line control plane so an agent can perform those operations through the running DeepChat main process without touching credentials or internal state directly.
Core Features & Use Cases
- Model Discovery and Invocation: List providers and models, inspect model configuration, and run raw provider calls with
model invoke using streaming JSONL output.
- Media and Document Operations: Generate images, video, and speech, transcribe audio artifacts, and run local OCR extraction with results returned as artifact metadata.
- Configuration and Management: Read settings, Skills, and MCP server lists, and request renderer-approved changes such as preference settings, HTTPS Skill installs, and new remote MCP configurations.
- Use Case: A user asks the agent to compare two models on a prompt. The agent lists providers and models, invokes each model with the same prompt via
deepchat model invoke --jsonl, and reports the structured results without mutating any global defaults.
Quick Start
Ask the agent to check DeepChat's runtime status and list enabled providers, then invoke a chosen model with your prompt using the deepchat CLI.
