Need Developer Sites to Be AI-Scrapable
Developer documentation sites must be rebuilt to enable AI agents to efficiently consume them, including markdown endpoints for full content access.
Take Control Of Your Signals — Become a Naftiko Design Partner Today!
Persona Story:
Nico, the partner AI lead, wants developer documentation sites rebuilt to enable AI agents to efficiently consume them—including markdown endpoints for full content access.
Problem Context
- MCP’s own dev site provides .md endpoint with entire documentation in markdown
- Most developer sites aren’t designed for AI consumption
- AI agents struggle to navigate complex documentation hierarchies
Problem Impact
- AI agents can’t efficiently learn from documentation
- Manual effort required to prepare documentation for AI consumption
- Competitive disadvantage vs. companies with AI-friendly docs
Naftiko Today
- MCP exposure (Streamable HTTP and stdio transports) makes capability definitions natively consumable by AI agents without scraping
- Agent Skills with auto-exposed HTTP APIs provide structured, machine-readable endpoints for skill discovery and download
- outputParameters with JSONPath normalization ensures AI agents receive clean, structured data rather than raw HTML or unstructured page content
- Executable YAML specs are inherently machine-readable, eliminating the need to reverse-engineer documentation into usable formats
Naftiko Tomorrow
- HTML/Markdown format support (Second Alpha) would let capabilities serve content in formats AI agents already consume natively
- A2A adapter (Second Alpha) would enable agent-to-agent discovery and communication, going beyond scraping to structured agent protocols
- Naftiko Shipyard MVP (Second Alpha) would provide a centralized, machine-queryable registry of all available capabilities across the organization
- Fabric capability discovery (v1.1) would enable cross-organization capability discovery for AI agents at scale