Firecrawl, the web-scraping and data-extraction API startup, released Developer Index, a purpose-built search layer that gives AI coding agents structured access to READMEs, external documentation, GitHub issues, pull requests, and OpenAPI specs across more than 70 million sources, with most content refreshed daily. The pitch is that general-purpose web search and generic RAG pipelines are a poor fit for the kind of lookups coding agents actually need, such as finding the right function signature in a specific library version, checking whether an open GitHub issue matches a bug the agent just hit, or pulling the exact OpenAPI schema for an endpoint, because that information lives in fragmented, fast-changing developer artifacts rather than well-indexed web pages. Alongside the index, Firecrawl published DevDex, an open benchmark of 1,179 real developer-search queries scored on Recall@10 and MRR@10, and reported that Developer Index beats the next-best external retrieval provider by roughly 10 percentage points on that benchmark. For teams building or evaluating coding agents, whether that's an internal tool, an IDE plugin, or a customer-facing product, this is a concrete data point in the increasingly crowded space of context providers for agentic coding: retrieval quality over code-specific sources is becoming its own specialized infrastructure layer, separate from general web search APIs. It's also a signal that benchmark transparency, publishing DevDex alongside the product, is becoming table stakes for anyone selling into the agent-tooling market, since buyers now expect a reproducible way to compare retrieval providers rather than relying on vendor claims alone.