Skip to main content
This page contains two types of examples to help you build agents with the Hive framework: Templates (ready-to-run code) and Recipes (design blueprints).

​
Recipes vs Templates

Recipes

“How to make it” - Prompt-only descriptions with goals, nodes, and routing logic. Not runnable code.

Templates

“Ready to eat” - Working agent scaffolds that follow the standard Hive export structure.

​
Templates

A template is a working agent scaffold that follows the standard Hive export structure. Copy the folder, rename it, swap in your own prompts and tools, and run it.

​
When to Use Templates

  • Get a new agent running quickly
  • Start from a known-good structure instead of from scratch
  • See how all the pieces (goal, nodes, edges, config, CLI) fit together in real code

​
Template Structure

Each template is a complete agent package: 
template_name/
├── __init__.py       # Package exports
├── __main__.py       # CLI entry point
├── agent.py          # Goal, edges, graph spec, agent class
├── agent.json        # Agent definition (used by build-from-template)
├── config.py         # Runtime configuration
├── nodes/
│   └── __init__.py   # Node definitions (NodeSpec instances)
└── README.md         # What this template demonstrates

​
How to Use a Template

Use the /hive-create skill and select “From a template” to interactively pick a template, customize the goal/nodes/graph, and export a new agent.

​
Available Templates

Deep Research Agent

Interactive research agent that searches diverse sources, evaluates findings with user checkpoints, and produces a cited HTML report.Features:
  • Multi-source research
  • User checkpoint validation
  • HTML report generation
  • Citation tracking
Usage:
PYTHONPATH=core:examples/templates python -m deep_research_agent run --topic "AI"

Email Inbox Management

Automatically manage Gmail inbox using user-defined rules. Fetch, classify, and take actions on emails.Features:
  • Rule-based email classification
  • Batch processing (configurable)
  • Multiple actions (trash, spam, archive, star)
  • Summary reporting
Nodes: intake → fetch-emails → classify-and-act → reportTools: gmail_list_messages, gmail_get_message, gmail_trash_message, gmail_modify_message

Job Hunter

Analyze resume, find matching jobs, and generate customized application materials.Features:
  • Resume analysis and role matching
  • Job search and scraping
  • User job selection
  • Custom resume suggestions
  • Cold outreach email generation
Nodes: intake → job-search → job-review → customizeTools: web_search, web_scrape, save_data

Tech News Reporter

Research the latest technology and AI news from the web and produce a well-organized report.Features:
  • Recent tech/AI news search
  • Multi-source scraping
  • Structured HTML report
  • Source attribution
Nodes: intake → research → compile-reportTools: web_search, web_scrape, save_data, serve_file_to_user

Competitive Intel Agent

Monitor competitors and generate intelligence reports.Features:
  • Competitor monitoring
  • Market analysis
  • Intelligence reporting

Vulnerability Assessment

Security scanning and vulnerability assessment agent.Features:
  • Security scanning
  • Vulnerability detection
  • Assessment reports

​
Recipes

A recipe describes an agent’s design — the goal, nodes, prompts, edge logic, and tools — without providing runnable code. Think of it as a blueprint: it tells you how to build the agent, but you do the building.

​
When to Use Recipes

  • Understand a pattern before committing to an implementation
  • Adapt an idea to your own codebase or tooling
  • Learn how to think about agent design (goals, nodes, edges, prompts)

​
What’s in a Recipe

Each recipe is a markdown file containing:
  • Goal: What the agent accomplishes, including success criteria and constraints
  • Nodes: Each step in the workflow, with the system prompt, node type, and input/output keys
  • Edges: How nodes connect, including conditions and routing logic
  • Tools: What external tools or MCP servers the agent needs
  • Usage notes: Tips, gotchas, and suggested variations

​
How to Use a Recipe

1

Read the recipe

Understand the design pattern and goals.
2

Create a new agent

Use the standard export structure (see templates for a scaffold).
3

Translate to code

Convert the recipe’s goal, nodes, and edges into implementation.
4

Wire in tools

Connect the tools described in the recipe.
5

Test and iterate

Run and refine your implementation.

​
Available Recipes

Social Media Management

Schedule posts, reply to comments, monitor trends

Newsletter Production

Transform voice memos and ideas into polished emails

News Jacking

Personalized outreach triggered by real-time company news

Ad Campaign Monitoring

Monitor and analyze advertising campaign performance

CRM Update

Ensure every lead has follow-up dates and status

Inquiry Triaging

Sort tire kickers from hot leads

Onboarding Assistance

Guide new clients through setup and welcome kits

Inbox Management

Clear spam and surface emails that need your brain

Invoicing & Collections

Send invoices and chase overdue payments

Data Keeper

Pull data from multiple sources into unified reports

Calendar Coordination

Protect Deep Work time and book travel

Quality Assurance

Test features and links before they go live

Documentation

Turn messy processes into clean SOPs

Support Troubleshooting

Handle Level 1 tech support

Issue Triaging

Categorize and route bug reports by severity

​
Getting Started

View All Templates

Browse the complete collection of working templates on GitHub

View All Recipes

Explore design blueprints for common agent patterns

Contributing

Learn how to contribute your own examples

Developer Guide

Complete guide for building agents

Build docs developers (and LLMs) love

Get started for freeTalk to us