Skip to main content
The Orthogonal skill file (skill.md) provides AI agents with everything they need to discover and use APIs - no integration required.

​
What is skill.md?

A skill file is a markdown document that teaches AI agents how to use a service. When an agent reads the Orthogonal skill file, it learns:
  • How to authenticate
  • Available tools (search, details, integrate, run)
  • Request/response formats
  • Example workflows

​
Fetching the Skill

Your agent can fetch the latest skill file from: 
https://orthogonal.com/skill.md
Or programmatically: 
curl https://orthogonal.com/skill.md

​
Using with AI Agents

​
Claude / Cursor / OpenClaw

Add Orthogonal to your agent’s skills directory, or have it fetch dynamically: 
# In your agent prompt

Before using external APIs, fetch the Orthogonal skill:
curl https://orthogonal.com/skill.md

​
Custom Agents

Inject the skill content into your agent’s context: 
const skillContent = await fetch('https://orthogonal.com/skill.md').then(r => r.text());

const messages = [
  { role: 'system', content: skillContent },
  { role: 'user', content: 'Find contact info for the CEO of Stripe' }
];

​
What’s in the Skill File

The skill file includes:

​
Authentication

export ORTHOGONAL_API_KEY=orth_live_your_api_key

​
Available Tools

ToolEndpointDescription
searchPOST /v1/searchFind APIs with natural language
get_detailsPOST /v1/detailsGet endpoint parameters
integratePOST /v1/integrateGet code snippets
usePOST /v1/runExecute API calls
list_allGET /v1/list-endpointsList all APIs

​
Typical Workflow

1. Search for what you need → POST /v1/search
2. Get details for the endpoint → POST /v1/details  
3. Use the endpoint → POST /v1/run

​
Code Examples

The skill includes curl examples for each tool, plus SDK and x402 integration patterns.

​
Benefits

Always Up-to-Date

Fetching dynamically ensures your agent has the latest API information.

No Integration

Agents can start using Orthogonal immediately just by reading the skill.

Self-Documenting

The skill teaches agents the full workflow, not just endpoints.

Universal

Works with any agent that can read markdown and make HTTP requests.

​
Example Agent Session

User: Find the CEO of Stripe and get their full contact info

Agent: *fetches skill.md*
Agent: *calls POST /v1/search with prompt "enrich lead find email"*
Agent: Found Sixtyfour API - enrich-lead, find-email, find-phone

Agent: *calls /enrich-lead* → Gets name, title, company
Agent: *calls /find-email* → Gets email address
Agent: *calls /find-phone* → Gets phone number

Here's what I found:
- Name: Patrick Collison
- Title: CEO at Stripe
- Email: [email protected]
- Phone: +1-xxx-xxx-xxxx
- LinkedIn: linkedin.com/in/patrickcollison