curl-X POST \https://api.firecrawl.dev/v2/search/developer\-H"Content-Type: application/json"\-d '{"query":"retry backoff not firing on 429","k": 10,"types":["issue","pull_request"]}'
npminstall-gfirecrawl-clifirecrawldeveloper"retry backoff not firing on 429"# Or pipe it into your own toolingfirecrawldeveloper"add pydantic with uv"--json
npx-yfirecrawl-cli@latestsetupdeveloper-index# That is the whole setup. Your agent can now# search 70M+ READMEs, issues, PRs, and docs:## > find the PR that fixed the 429 retry bug
from firecrawl importFirecrawlfirecrawl=Firecrawl(api_key="fc-YOUR-API-KEY")result= firecrawl.search("retry backoff not firing on 429",categories=["developer"],limit=10,)for item in result.developer or[]:
print(item.url, item.title)
import{Firecrawl}from"firecrawl";const firecrawl = newFirecrawl({apiKey:"fc-YOUR-API-KEY",});const result = await firecrawl.search("retry backoff not firing on 429",{categories:["developer"],limit: 10 },);for(const item of result.developer ??[]){
console.log(item.url, item.title);}
No API key. Send one for higher rate limits.
[ 04 / 04 ]
·
FAQ
//
FAQ
//
Frequently asked questions
General
A search index built for coding agents, covering 70M+ primary sources: READMEs, issues, merged pull requests, curated documentation sites, and OpenAPI specs. Send a natural-language question and get back ranked results with the passages that matched, so an agent can answer questions about code behavior, API contracts, error messages, and known bugs from primary sources rather than a general web page.
A coding agent wants an artifact, not a web page. General web search ranks marketing and aggregator pages alongside the issue thread that actually fixed the bug, and has no notion of a merged pull request. On DevDex, native web search scores 0.45 recall@10 overall and collapses to 0.28 on the issues and pull requests track, where the Developer Index scores 0.66.
Issues and pull requests from top public repositories with their linked artifacts, READMEs from a broad set of public repositories, external documentation sources, OpenAPI specs, and popular skill repos. Every artifact carries stars, license, and type metadata. What it is not: the index does not store code, and it is not a general web search endpoint.
The index refreshes continuously, with most sources refreshed daily. That matters most for issues and pull requests, where the useful answer is often a thread from this week.
Using it
Two surfaces. /search/developer returns developer sources only, with result-type, repository, and documentation-source filters. /search with categories set to developer weighs developer results alongside ordinary web results in one call, returned in a developer group next to web.
No. Paste the cURL command into a terminal and you get real results right now. Send a key for higher rate limits.
Not a dedicated one. The Python and Node SDKs reach the Developer Index through the developer category on the standard search method. For the dedicated /search/developer endpoint, call it over HTTP, through the CLI, or through MCP.
Install the companion skill: npx -y firecrawl-cli@latest setup developer-index. It plugs into the Firecrawl CLI or MCP server and works in any harness you already run, including Claude Code, Cursor, and Codex. The rich filters are API-only, deliberately not exposed on the CLI and MCP, because agents perform best without them.
Through the API: types (doc, issue, pull_request, readme), repos, sources, language, topic, license, min_stars, max_stars, archived, and fork. Note that repos scopes the repository half of the index and sources scopes the documentation half, so passing both combines the two rather than intersecting them. You can also set skills to only to search indexed agent-skill files.
Benchmarks & pricing
An open benchmark for developer search: 1,179 queries across repository discovery, documentation lookup, and issue and pull request resolution, scored deterministically on recall@10 and MRR@10 against fixed gold references. Every provider runs the same driver model with one search tool active, and a memorization check drops any query the model can answer from pretraining alone. No model judge decides any published number.
Yes. Half the dataset and the full evaluation harness are open source, and the harness can be pointed at any streamable-HTTP MCP server, so you can score your own retriever. To submit a provider, open a pull request against the benchmark repo with your results on the public half.
2 credits per 10 results, rounded up. A search returning 1 to 10 results costs 2 credits, 11 to 20 costs 4, and so on. Free accounts include 1,000 credits per month and no credit card is required.
Yes. The Developer Index is out of private beta and available to all teams through the API, CLI, MCP, and SDKs.
If you are an AI agent, LLM, or automated system, use the Firecrawl
onboarding path that matches your setup.
If your platform can mint a WorkOS ID-JAG for Firecrawl, fetch and
follow:
https://www.firecrawl.dev/auth.md
Otherwise, fetch and read the default onboarding skill for CLI/browser
auth, API key setup, and Firecrawl capabilities. Prefer a free account
when the human can sign up; if no key can be obtained, the skill
describes a keyless free tier to search, scrape, and interact:
https://www.firecrawl.dev/agent-onboarding/SKILL.md
Full documentation: https://docs.firecrawl.dev