Skip to content

Examples

Real-world agent specifications demonstrating different use cases and patterns.

AI Engineer Agent

The included AI Engineer example demonstrates a comprehensive agent specification with MCP integrations.

View the Specification

Bash
1
2
3
4
5
# View the Markdown specification
uv run agent-script-markdown

# Or view the file directly
cat examples/ai-engineer.md
Bash
1
2
3
4
5
# View the JSON specification
uv run agent-script-json

# Or view the file directly
cat examples/ai-engineer.json

Key Features Demonstrated

  • Multi-Phase Interaction

    Three-phase workflow from requirements to implementation with structured steps

  • MCP Integration

    Context7 and sequential-thinking servers for research and complex reasoning

  • Behavioral Rules

    Clear DO and DO-NOT guidelines for secure, production-ready development

  • Deliverables

    Specific output requirements including code, tests, and documentation

Try the Examples

Convert Between Formats

Bash
1
2
3
4
5
# Convert JSON to Markdown
uv run convert-to-markdown examples/ai-engineer.json -o my-agent.md

# Convert Markdown to JSON  
uv run convert-to-json examples/ai-engineer.md -o my-agent.json

Validate and Test

Bash
1
2
3
4
5
# Generate JSON schema
uv run generate-schema

# View the schema
uv run show-spec-json

Creating Your Own

Use the AI Engineer specification as a template:

  1. Copy the structure from examples/ai-engineer.md
  2. Customize the identity (name, role, expertise)
  3. Define your capabilities and MCP integrations
  4. Design your interaction model phases and steps
  5. Set behavioral rules and expected deliverables
  6. Validate with the conversion tools

Next Steps