Learn by Example
Practical examples and tutorials to help you build autonomous agent workflows with Radium. Filter by category and difficulty to find exactly what you need.
Showing 8 of 8 examples
Multi-Agent Code Review Workflow
Orchestrate multiple agents to analyze, review, and improve code. Demonstrates DAG workflows, agent coordination, and error handling with automatic retries.
# workflow.toml
[[agents]]
id = "analyzer"
model = "claude-3-sonnet"
[[agents]]
id = "reviewer"
model = "gpt-4"
[workflow]
steps = [
{ agent = "analyzer", input = "task.code" },
{ agent = "reviewer", depends_on = ["analyzer"] }
]Self-Hosted AI with Ollama
Run agents completely locally using Ollama. No API keys required - perfect for privacy-sensitive applications or offline development. Supports CodeLlama, Mistral, and more.
# agents/local-coder.toml
[agent]
id = "local-coder"
name = "Local Code Assistant"
[model]
provider = "ollama"
model = "codellama:34b"
base_url = "http://localhost:11434"Cost-Optimized Agent Selection
Automatically select the most cost-effective model for each task. Uses policy-driven orchestration to balance cost, speed, and quality based on your requirements.
[orchestration]
policy = "cost-optimized"
max_cost_per_task = 0.50
[agent_selection]
prefer_local = true
fallback_strategy = "cheapest-cloud"
[models.preference]
fast_tasks = ["gpt-3.5-turbo", "claude-haiku"]
complex_tasks = ["gpt-4", "claude-sonnet"]Autonomous Debugging Agent
Enable agents to autonomously debug failing tests. Demonstrates YOLO mode, bounded autonomy, and intelligent error recovery with minimal human intervention.
[execution]
mode = "autonomous"
require_approval = false
max_iterations = 10
budget = 5.00
[oversight]
vibe_check_frequency = "every-iteration"
on_anomaly = "pause-and-notify"CI/CD Pipeline Integration
Integrate Radium agents into your CI/CD pipeline. Automate code reviews, documentation generation, and test creation as part of your build process.
# .github/workflows/ai-review.yml
- name: Run AI Code Review
run: |
radium-cli chat code-reviewer \
"Review PR #${{ github.event.number }} \
--mode autonomous \
--max-cost 2.00Building Custom Extensions
Create custom tools and integrations for your agents. This example shows how to build a GitHub extension that lets agents search issues, create PRs, and manage repositories.
use radium_core::Extension;
#[derive(Extension)]
pub struct GitHubExtension {
api_key: String,
}
impl GitHubExtension {
fn search_issues(&self, query: &str) -> Result<Vec<Issue>> {
// Implementation
}
}Custom Agent Personas
Define custom personas to shape agent behavior and expertise. This example creates a security-focused reviewer with specific knowledge domains and communication style.
# personas/security-expert.toml
[persona]
id = "security-expert"
expertise = ["security", "owasp", "cryptography"]
style = "thorough"
tone = "professional"
[prompts]
system = """
You are a cybersecurity expert specializing in
application security and the OWASP Top 10...
"""ETL Data Pipeline Orchestration
Build intelligent ETL pipelines with AI-powered data validation and transformation. Agents can detect anomalies, suggest optimizations, and handle errors autonomously.
[[workflow.steps]]
id = "extract"
agent = "data-extractor"
[[workflow.steps]]
id = "validate"
agent = "data-validator"
depends_on = ["extract"]
[[workflow.steps]]
id = "transform"
agent = "data-transformer"
depends_on = ["validate"]
parallel = ["quality-check"]Ready to Build?
Install Radium and start building your own autonomous agent workflows. Check out our comprehensive documentation for detailed guides and API references.