Table of Contents

Quickstart

Get productive with JD.AI in under five minutes. This guide walks through a typical workflow — from launching the assistant to committing changes — so you can see how every major feature fits together.

Quickstart terminal workflow

Before you begin

Make sure you have:

  • JD.AI installed — dotnet tool install -g JD.AI
  • At least one provider configured — Claude Code, GitHub Copilot, OpenAI Codex, Ollama, or a local GGUF model

Quick provider setup — the fastest way to get running with an API key provider:

# Set your API key as an environment variable, then launch
export OPENAI_API_KEY=sk-...
jdai --provider openai

This works for any key-based provider — just swap the variable and flag (e.g. ANTHROPIC_API_KEY + --provider anthropic).

See Installation for full installation instructions and Provider Setup for all provider options.

Step 1: Start JD.AI in your project

cd /path/to/your/project
jdai

On startup JD.AI detects every available provider, selects the best model, and displays a welcome banner with provider/model, service health, current working directory, and version details. You can customize what appears with /config set welcome_* ... (for example /config set welcome_cwd off) and optionally enable MoTD via welcome_motd + welcome_motd_url.

Step 2: Find and select a model

Use /model search to discover available models across all configured providers:

/model search gpt-4
/model search claude
/model search llama

Select a result to switch to it immediately. You can also pull new local models this way.

Step 3: Explore your codebase

Ask plain-language questions to orient yourself:

what does this project do?
explain the folder structure
where is the main entry point?

JD.AI reads files, follows references, and synthesizes an answer — no manual searching required.

Step 4: Make a code change

Describe the change you want:

add input validation to the user registration form

JD.AI locates the relevant files, proposes edits, and asks you to confirm before applying them. Each tool invocation (file read, file write, shell command) requires your approval unless you have opted into auto-run.

Step 5: Use tools

JD.AI has built-in developer tools — file I/O, grep, shell, git, and web search. Ask naturally and the right tool is selected automatically:

search for all TODO comments in the codebase

The assistant invokes the grep tool, streams results back, and summarizes the findings.

Step 6: Work with git

what files have I changed?
commit my changes with a descriptive message

JD.AI runs git status and git diff, drafts a commit message, and executes the commit after your approval.

Step 7: Spawn a subagent

For larger tasks you can delegate to a scoped subagent:

use an explore agent to find how authentication works in this codebase

Subagents run in their own context with scoped tool access and report results back to the main conversation.

Step 8: Switch providers mid-session

You can change providers or models at any point without losing your conversation:

/provider            # Show current provider
/provider anthropic  # Switch to Anthropic
/model gpt-4.1       # Switch to a specific model
/providers           # List all available providers with status

Your conversation history carries over — the new model picks up where the previous one left off.

Step 9: Use slash commands

Key commands to know:

  • /help — list all commands
  • /model <name> — switch model
  • /compact — compress conversation history to free up context
  • /save — persist the current session
  • /sessions — list saved sessions

Step 10: Save and resume sessions

Name and save your session so you can pick up later:

/name my-feature-session
/save

Resume it in a future run:

/sessions
/resume <id>

Essential commands

Command What it does
jdai Start interactive mode
/help Show available commands
/models List available models
/model search <query> Search for models across providers
/default provider <name> Set the default provider
/default model <name> Set the default model
/provider Show or switch the active provider
/mcp Manage MCP server connections
/workflow Run a saved workflow
/plan Create or view an execution plan
/fork Fork the conversation into a new branch
/compact Compress conversation
/save Save current session
/quit Exit JD.AI

Pro tips

  • Run /compact before your context window fills up — it summarizes the conversation and reclaims token space.
  • Spawn subagents for specialized tasks like code review, exploration, or multi-file refactoring.
  • Create a JDAI.md file in your repository root with project-specific instructions that JD.AI reads on startup.
  • Use /autorun to skip tool-confirmation prompts during repetitive workflows.

What's next

Edit this page