Fetching from the wire…
01
02
03
04
05
06
07
08
Public story · 2026-08-03 · high
The single-binary tool skips per-request browser boots by running a V8 isolate in-process, hitting standard pages in about 300ms.
Why now: As of August 3, Draco sits at 48 stars, an early test of whether self-hosted scraping beats paying Firecrawl per request.
Draco reimplements Firecrawl's REST API in a single Rust binary you self-host instead of paying per scrape, according to its GitHub repo.
Per-request pricing is a real line item in agent pipelines that scrape at volume. Draco's pitch: make that line item disappear once you self-host.
Firecrawl and similar tools typically boot a full browser to render JavaScript-heavy pages before scraping them. Draco runs a V8 isolate in-process instead, and the author cites roughly 300ms for standard HTML pages.
It extracts Markdown and metadata natively. For structured data it uses a tiered approach: embedded state first, then API discovery, then runtime interception if neither exists.
It also runs multi-engine web search, supports MCP, and keeps interactive sessions with cookie persistence. Licensing is dual MIT and Apache-2.0.
Yes, but Firecrawl has 159,000 stars to Draco's 48, and matching an API surface on paper isn't the same as proving it at Firecrawl's scale.
Each link below shares sources, entities, or timing with this story.
Firecrawl supports Codex / Shared entities / Same source domain / Shared topic / Earlier coverage
Linked by a graph relationship (Firecrawl supports Codex); both cover Apache, MCP, MIT; reported by the same outlet (github.com).
Linked by a graph relationship (Firecrawl supports Codex); both cover Apache, HTML; reported by the same outlet (github.com).
Firecrawl supports Claude Code / Shared entities / Same source domain / Earlier coverage
Linked by a graph relationship (Firecrawl supports Claude Code); both cover Apache, HTML, MCP; reported by the same outlet (github.com).
Firecrawl supports Codex / Shared entities / Same source domain / Earlier coverage
Linked by a graph relationship (Firecrawl supports Codex); both cover MCP, MIT, Rust; reported by the same outlet (github.com).
Google released MCP / Shared entities / Same source domain / Earlier coverage
Linked by a graph relationship (Google released MCP); both cover MCP, REST API, Rust; reported by the same outlet (github.com).
Firecrawl supports Codex / Shared entities / Same source domain / Shared topic / Earlier coverage
Linked by a graph relationship (Firecrawl supports Codex); both cover MCP, Rust; reported by the same outlet (github.com).
Linked by a graph relationship (Firecrawl supports Codex); both cover MCP, Rust; reported by the same outlet (github.com).
Firecrawl supports Claude Code / Shared entity: MCP / Same source domain / Shared topic / Earlier coverage / Tension
Linked by a graph relationship (Firecrawl supports Claude Code); both cover MCP; reported by the same outlet (github.com).