Distribution

Can agents find and select your API?

Before an agent can use your API, it must know your API exists. Distribution is about how agents discover, evaluate, and select your API from all available options. In a world where tool selection is increasingly automated, being present in the right registries and providing clear signals about your capabilities is as important as the capabilities themselves.

Key principles

  • Publish to package registries so agents can install your SDK without manual configuration
  • Make your documentation crawlable and machine-readable
  • Expose discovery surfaces like llms.txt and sitemap.xml
  • Give agents clear signals about what your API does and when to use it
  • Be present where agents search: registries, directories, and aggregators

Channel distribution

Agents discover APIs through registries, marketplaces, and directories. Unlike human developers who might find an API through a blog post or conference talk, agents rely on structured registries where tools are cataloged with machine-readable metadata.

Every registry you publish to is another path an agent can take to find you.

  • SDK package registries - npm, PyPI, RubyGems, NuGet, and others relevant to your audience
  • MCP registries - register your MCP servers so tool-calling agents can discover them
  • Tool directories and plugin marketplaces - skill registries and capability catalogs
  • Documentation aggregators - services like Context7 that index API documentation and make it available to agents in standardized formats

Content

The content you publish affects whether agents recommend your API when solving problems. All of the following contribute to the training data and retrieval corpora that inform agent behavior:

  • Documentation sites
  • Blog posts and guides
  • Product pages
  • GitHub repositories
  • Community forums

Describe what problems your API solves and when it is the right choice.

Agents evaluating multiple APIs for a task need clear signals about fit, not just feature lists.

Search optimization

Traditional SEO still matters, but agents also use structured data and machine-readable formats to evaluate APIs.

Structured data

  • JSON-LD markup helps agents understand your API's capabilities and relationships
  • Structured FAQ sections provide direct answers that agents can extract and present

Generative search

  • GEO (Generative Engine Optimization) focuses on making content surfaceable by AI-powered search and recommendation systems
  • AEO (Answer Engine Optimization) emphasizes clear factual claims and direct answers over keyword optimization

Dynamic discovery

Expose machine-readable discovery surfaces so agents can understand your API without reading documentation intended for humans.

Discovery files

llms.txt at your domain root provides a concise, agent-friendly overview of what your API does and how to get started.

sitemap.xml helps agents navigate your documentation.

robots.txt with permissive rules ensures agents can actually access the content they find.

These files cost almost nothing to create but significantly improve the likelihood that an agent will correctly identify and select your API for a task.