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.

shob agent lets you define custom AI agents with tailored system prompts, tool sets, models, and operating modes. A custom agent is stored as a Markdown file with YAML frontmatter and can be used in place of the default agent by passing --agent <name> to shob run.

Usage

shob agent <subcommand>

shob agent create

Launches a wizard (or runs fully non-interactively when all flags are provided) to generate a new agent configuration file. Shob uses the provided description to generate a system prompt and an identifier for the agent using the configured AI model.
shob agent create [flags]

Flags

--path
string
Directory path where the agent file should be written. If omitted, the wizard asks whether to save the agent to the current project (.shob/agent/) or globally (~/.config/shob/agent/).
--description
string
A plain-language description of what the agent should do. Shob passes this to the AI model to generate the system prompt. If omitted, the wizard prompts for it interactively.
--mode
string
The role this agent can fulfil. Accepted values:
  • all — the agent can act as both a primary agent and a subagent (default when using the wizard)
  • primary — the agent acts as the main, top-level agent
  • subagent — the agent can only be invoked by another agent as a subtask
--tools
string
Comma-separated list of tools to enable for the agent. Pass an empty string ("") to enable all tools. If omitted, the wizard shows a multi-select prompt with all tools pre-selected.Available tools: bash, read, write, edit, list, glob, grep, webfetch, task, todowrite
--model
string
Alias: -mThe model to use when generating the agent’s system prompt, in provider/model format. Defaults to the project’s configured model.

Non-interactive example

When --path, --description, --mode, and --tools are all provided, shob agent create runs without any prompts and prints the created file path to stdout.
shob agent create \
  --path .shob/agent \
  --description "Reviews pull requests and suggests improvements to code quality and test coverage" \
  --mode primary \
  --tools "bash,read,glob,grep,webfetch"

Interactive example

shob agent create
The wizard will ask for:
  1. Location (current project or global)
  2. Description of the agent
  3. Which tools to enable
  4. Agent mode (all, primary, or subagent)
The generated agent file is a Markdown file with YAML frontmatter, for example:
---
description: "When to use this agent — generated from your description"
mode: all
tools:
  todowrite: false
---

You are a code review assistant...

shob agent list

Lists all available agents — both built-in native agents and custom agents — sorted with native agents first, then alphabetically by name.
shob agent list

Example output

coder (all)
  {"rules":[],"permissions":[]}
reviewer (primary)
  {"rules":[],"permissions":[]}
my-custom-agent (subagent)
  {"rules":[],"permissions":[]}
Each entry shows the agent name, its mode in parentheses, and its permission configuration as JSON.

Using a custom agent

Pass the agent name to shob run with --agent:
shob run --agent reviewer "review the changes in the last commit"

Build docs developers (and LLMs) love