Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/shobcoder/shob/llms.txt

Use this file to discover all available pages before exploring further.

By the end of this guide you will have Shob installed, an AI provider authenticated, and your first coding task running in your terminal. The whole process takes under five minutes on a standard internet connection.
1

Install Shob

Install the shob-code package globally using your preferred package manager.
npm install -g shob-code
Verify the installation succeeded by checking the version:
shob --version
You should see output similar to:
0.0.2
2

Authenticate a Provider

Shob needs access to an AI provider to power its agents. Run the interactive login flow to add your first credential:
shob providers login
The CLI presents an interactive provider picker. Use the arrow keys to select your provider, then press Enter:
┌  Add credential

◆  Select provider
│  ● Anthropic
│  ○ OpenAI  (ChatGPT Plus/Pro or API key)
│  ○ Google
│  ○ OpenRouter
│  ○ Vercel
│  ○ Other

After choosing a provider you will be prompted for your API key:
┌  Add credential

◇  Select provider
│  Anthropic

◆  Enter your API key
│  ▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪▪

Press Enter after pasting your key. You will see a confirmation:
◇  Login successful

└  Done
If you already know your provider ID and want to skip the interactive picker, pass it directly:
shob providers login --provider anthropic
To see all configured credentials at any time, run:
shob providers list
3

Run Your First Task

Navigate to your project directory and send a message to Shob using shob run:
cd your-project
shob run "Add input validation to the login form and write a unit test for it"
Shob starts a new session, picks an available model from your configured provider, and begins planning the task:
   _____ _           _
  / ____| |         | |
 | (___ | |__   ___ | |__
  \___ \| '_ \ / _ \| '_ \
  ____) | | | | (_) | |_) |
 |_____/|_| |_|\___/|_.__/

Session abc123 · Anthropic · claude-sonnet-4-5
──────────────────────────────────────────────
✦  Planning task...
✦  Reading src/components/LoginForm.tsx
✦  Reading src/tests/LoginForm.test.tsx
You can pin a specific model with the --model flag, for example --model anthropic/claude-opus-4-5. The value should be in provider/model format.
4

Review and Approve Permissions

Before Shob writes any file or runs any command it will pause and ask for your approval:
┌  Permission request

│  Write file
│  src/components/LoginForm.tsx

◆  Approve?
│  ● Yes, allow once
│  ○ Yes, allow for this session
│  ○ No, deny

Choose Yes, allow for this session to approve similar actions for the rest of this task without being prompted again, or Yes, allow once to review each action individually.
The --dangerously-skip-permissions flag on shob run will auto-approve all non-denied permissions without prompting. Use it only in trusted automated environments after carefully auditing your permission configuration.
5

View Results and Next Steps

Once the task is complete, Shob prints a summary of everything it changed:
✦  Modified  src/components/LoginForm.tsx
✦  Created   src/tests/LoginForm.validation.test.tsx
✦  Session complete · 2 files changed

Resume with: shob run --continue
You can inspect the diff in your editor, run your test suite, and then continue the conversation in the same session:
# Continue the last session
shob run --continue "Also add server-side validation in the API route"

# Continue a specific session by ID
shob run --session abc123 "Update the error messages to match our design system"
Attach files to your message with -f to give the agent additional context without pasting code into the prompt:
shob run -f design-spec.md "Implement the button component shown in this spec"

Sessions

Learn how sessions work, how to resume them, and how to fork them for safe experimentation.

Configuration

Customise providers, models, permission rules, and plugins in shob.json.

run command

Full reference for all flags available on shob run.

Build docs developers (and LLMs) love