AI Providers

JSAT supports six AI backends. The right one is selected automatically based on what is available, but you can override it at any time.

Provider Overview

| Bob Shell CLI | jsat ai use bob | Free tier | bob binary |

Provider	Switch command	Cost	Needs
Claude Code CLI	jsat ai use claude-cli	Free tier	claude binary
Anthropic API	jsat ai use anthropic	Paid	ANTHROPIC_API_KEY
OpenAI	jsat ai use openai	Paid	OPENAI_API_KEY
Google Gemini	jsat ai use gemini	Paid	GEMINI_API_KEY
Ollama (local)	jsat ai use ollama	Free	ollama serve + model
LM Studio (local)	jsat ai use lmstudio	Free	LM Studio running

---

Check What Is Available

jsat ai status

Example output:

        JSAT AI Providers
┌──────────────────────┬──────────────┬──────┬────────────────────────┐
│ Provider             │ Status       │ Free │ Notes / Models         │
├──────────────────────┼──────────────┼──────┼────────────────────────┤
│ claude (active)      │ ✓ available  │ yes  │ claude binary found     │
│ anthropic            │ ✓ key set    │ no   │ claude-sonnet-4-6      │
│ ollama               │ ✓ running    │ yes  │ llama3.2, phi3:mini    │
│ openai               │ ✗ no key     │ no   │ set OPENAI_API_KEY     │
│ lmstudio             │ ✗ not running│ yes  │ not running            │
└──────────────────────┴──────────────┴──────┴────────────────────────┘

Currently configured: claude_cli / claude-sonnet-4-6

---

Claude Code CLI (Recommended)

The Claude Code CLI (claude binary) is the default provider when it is installed. No API key is required — JSAT calls claude as a subprocess.

Install

# Follow instructions at claude.ai/code
# Verify:
which claude
claude --version

Activate

JSAT auto-detects the claude binary. No configuration needed. To set it explicitly:

jsat ai use claude-cli

This sets:

ai:
  provider: claude_cli
  model: claude-sonnet-4-6

Verify

jsat ai test

Inside the shell

> switch claude-cli

---

Anthropic API

---

Bob Shell CLI

Bob Shell is an IBM AI-powered terminal assistant that provides interactive coding assistance with multiple modes.

Install

npm install -g @ibm/bob-shell
# Verify:
which bob
bob --version

Activate

JSAT auto-detects the bob binary. No configuration needed. To set it explicitly:

jsat ai use bob

This sets:

ai:
  provider: bob_cli
  model: premium
  chat_mode: advanced

Available Modes

Bob Shell supports four interaction modes:

• plan — Planning and design mode
• code — Code implementation mode
• advanced — Advanced code mode with more tools (default)
• ask — Question and answer mode

Verify

jsat ai test

Inside the shell

> switch bob
> switch bob-cli

Open a Bob Shell session directly

jsat bob
jsat bob --mode advanced
jsat bob --resume <session-id>

Use the Anthropic API directly (requires a paid API key).

Install

pip install jsat[anthropic]
# or: pip install anthropic

Set API key

export ANTHROPIC_API_KEY=sk-ant-...
# Add to ~/.zshrc or ~/.bashrc to persist

Activate

jsat ai use anthropic

Available models: claude-sonnet-4-6 (default), claude-haiku-4-5-20251001, claude-opus-4-8.

Use a specific model:

jsat ai use anthropic --model claude-haiku-4-5-20251001

Verify

jsat ai test "say hello"

Inside the shell

> switch claude-api
> switch haiku    # shorthand for claude-haiku
> switch opus

---

OpenAI

Use GPT-4o or GPT-4o Mini.

Install

pip install jsat[openai]
# or: pip install openai

Set API key

export OPENAI_API_KEY=sk-...

Activate

jsat ai use openai
jsat ai use openai --model gpt-4o-mini   # cheaper

Default model: gpt-4o-mini.

Verify

jsat ai test

Inside the shell

> switch gpt
> switch gpt4mini

Open a GPT session directly

jsat gpt

---

Google Gemini

JSAT connects to Gemini via its OpenAI-compatible endpoint.

Set API key

Get a key from aistudio.google.com:

export GEMINI_API_KEY=...
# Also accepted: GOOGLE_API_KEY

Activate

jsat ai use gemini

This sets the provider to openai_compat with base_url pointed at:

https://generativelanguage.googleapis.com/v1beta/openai

Available models: gemini-1.5-flash (default, fast), gemini-1.5-pro (higher quality).

jsat ai use gemini --model gemini-1.5-pro

Verify

jsat ai test "what is 2 + 2?"

Inside the shell

> switch gemini
> switch gemini-pro

---

Ollama (Local, Free)

Ollama runs open-weight models locally. No API key, no data sent externally.

Install

macOSLinuxWindows

brew install ollama

curl -fsSL https://ollama.ai/install.sh | sh

Download the installer from ollama.ai.

Pull a model

ollama pull llama3.2       # 4 GB — good general quality
ollama pull phi3:mini      # 2 GB — fast, lower RAM usage
ollama pull qwen2.5-coder:7b  # 4 GB — code-focused

Start the server

ollama serve

JSAT checks http://localhost:11434 on startup. If Ollama is running, it is auto-selected when no cloud provider is configured.

Activate

jsat ai use ollama
jsat ai use ollama --model phi3:mini

List pulled models

jsat ai models

Verify

jsat ai test

Open an Ollama session

jsat ollama
jsat ollama --model phi3:mini

Inside the shell

> switch ollama
> switch phi    # shorthand for phi3:mini
> switch llama  # shorthand for llama3.2

---

LM Studio (Local, Free)

LM Studio lets you download and run models from Hugging Face with a local OpenAI-compatible API.

Install

Download from lmstudio.ai.

Load a model in LM Studio, then start the local server (usually at http://localhost:1234).

Activate

jsat ai use lmstudio

This sets the provider to openai_compat with base_url: http://localhost:1234/v1. JSAT will use whatever model LM Studio has loaded.

List loaded models

jsat ai models

Verify

jsat ai test

Inside the shell

> switch lmstudio

---

Auto-Detection Priority

When no provider is explicitly configured, JSAT probes all backends on startup and picks the best available one in this priority order:

1. Claude Code CLI (claude binary found on PATH)
2. Bob Shell CLI (bob binary found on PATH)
3. Anthropic API (ANTHROPIC_API_KEY set)
4. OpenAI API (OPENAI_API_KEY set)
5. Google Gemini (GEMINI_API_KEY or GOOGLE_API_KEY set)
6. Ollama (reachable at localhost:11434)
7. LM Studio (reachable at localhost:1234)

If none are reachable, JSAT runs without AI (graph queries only, no natural language).

---

Switching Providers

From the CLI

Update .jsat/config.yaml and test:

jsat ai use ollama
jsat ai test

From inside the JSAT shell

> switch claude
> switch claude-api
> switch claude-cli
> switch bob
> switch bob-cli
> switch gpt
> switch gpt4mini
> switch ollama
> switch phi
> switch llama
> switch gemini
> switch gemini-pro
> switch haiku
> switch opus
> switch lmstudio

With the Python SDK

from jsat import JSAT

js = JSAT(repo=".")
js.switch_ai("ollama", model="phi3:mini")

# Or set at construction
js = JSAT(repo=".", ai_provider="anthropic", model="claude-haiku-4-5-20251001")

---

Provider Configuration Reference

All AI settings live under the ai: key in .jsat/config.yaml:

ai:
  provider: ollama          # ollama | anthropic | openai | openai_compat | none
  model: llama3.2
  base_url: null            # set for lmstudio / gemini / custom endpoints
  max_tokens: 8192
  temperature: 0.1
  timeout_seconds: 120
  retry_attempts: 3

For Gemini:

ai:
  provider: openai_compat
  model: gemini-1.5-flash
  base_url: https://generativelanguage.googleapis.com/v1beta/openai

For LM Studio:

ai:
  provider: openai_compat
  model: local-model
  base_url: http://localhost:1234/v1

The ANTHROPIC_API_KEY, OPENAI_API_KEY, GEMINI_API_KEY environment variables are read at runtime. Do not put API keys in config.yaml.