firecrawl-knowledge-base Skill for Claude: What It Does and How to Install It

Makes your agent turn a docs site or a topic into organized, LLM-ready markdown, RAG chunks or a training file.

Installs: 32.7K; on skills.sh
Stars: 214; on GitHub
Updated: Nov 2026; last repo commit

This skill makes your agent collect web content and save it as files. It maps a documentation site, or searches a topic, scrapes each page to markdown, and keeps code examples and tables.

You pick one of four output modes: reference, RAG, training or docs mirror. It needs a Firecrawl API key and the Firecrawl CLI, and it writes into a .firecrawl/ folder in your project.

Good for

Developers who want local reference docs for an agent
Teams building a RAG corpus from public documentation
People preparing fine-tuning data from web sources

Install firecrawl-knowledge-base

Claude Code

Copy this prompt.

Install the firecrawl-knowledge-base skill in Claude Code. Run this command:

/plugin marketplace add firecrawl/firecrawl-workflows
/plugin install firecrawl-workflows@firecrawl-workflows

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Pick one route. The plugin routes install the whole firecrawl-workflows plugin: 16 skills (counted from the skills list in its plugin.json), and the repo tree has no hooks, agents or MCP config files. The skills-only routes install just firecrawl-knowledge-base. The skill drives the Firecrawl CLI and needs a Firecrawl API key. Install the Firecrawl CLI first, since the skill runs firecrawl commands: npm install -g firecrawl-cli (Firecrawl CLI docs), or npx -y firecrawl-cli@latest init -y --browser to install it, sign in and add the shared skills in one step. Then run firecrawl login or export FIRECRAWL_API_KEY. Who pays: the Firecrawl account that owns the key. Every call uses credits; a new account starts on the free plan with 1,000 credits a month and no card. Per Firecrawl's billing docs: map 1 credit per call; search 2 credits per 10 results; scrape 1 credit per page.

Built from the package data we hold. Not yet tested by us. Source.

What's in the skill

When it loads

Build a knowledge base from web content with Firecrawl. Use for local reference docs, RAG-ready chunks, fine-tuning datasets, documentation mirrors, topic corpora, or LLM-ready markdown organized from web sources.

Sections in SKILL.md

Onboarding Interview
Firecrawl Collection Plan
Parallel Work
Output Modes
Final Deliverable
Summary
Output Structure
Coverage
Usage Notes
Sources
Rerun Inputs
Quality Bar

Key facts

Author: Firecrawl (frontmatter author, and the repo owner)
License: ISC (frontmatter and the README)
Installs on skills.sh: 32,690 (Nov 2, 2026)
GitHub stars: 214 for the whole firecrawl/firecrawl-workflows repo (Nov 2, 2026)
Last repo commit: Nov 1, 2026
What it includes: 1 file: SKILL.md (bundles no scripts)
Needs: A Firecrawl API key (FIRECRAWL_API_KEY) with credits, and the Firecrawl CLI (npm install -g firecrawl-cli)
Part of: The firecrawl-workflows plugin: 16 skills in its Claude plugin.json (counted), 19 skill folders in the repo tree
Works in: AI coding agents, per the README. The repo also ships plugin manifests for Claude Code, Codex and Cursor

What it includes

Maps a documentation site, or searches for a topic-based corpus
The collection plan uses Firecrawl map for docs sites and search for topics.
Network · Map or search
Saves pages as .firecrawl/<hostname>/<path>/index.md
Preserves code examples and tables and removes navigation boilerplate where possible. Source URLs go into frontmatter or metadata.
Write · Scrape to markdown
Markdown files, an index.md and a sources.json
For local reference docs.
Write · Reference mode
Markdown files plus chunk files and a manifest.json
For retrieval pipelines.
Write · RAG mode
Scraped sources plus training-data.jsonl and training-metadata.json
SKILL.md asks for the training format if you want this mode.
Write · Training mode
A complete markdown mirror with a table of contents
For keeping a full local copy of a documentation site.
Write · Docs mirror mode
Splits work by docs section, source type, or scrape vs chunk vs manifest
Only when your agent supports sub-agents.
Runs code · Parallel splits

From the official docs

Permissions

Low risk

It sends the sites you name through Firecrawl using your API key and writes the pages it collects into a .firecrawl/ folder.

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

It writes a file tree into your project
SKILL.md saves each page under .firecrawl/<hostname>/<path>/index.md and adds index, chunk, manifest or JSONL files depending on the mode. Add .firecrawl/ to .gitignore if needed. github.com
Deep crawls make many Firecrawl calls
The rerun inputs include a depth of quick, thorough or exhaustive. Start with quick on a large site, since each page is a call under your key (our advice). github.com
It spends your Firecrawl credits
Every call is billed in credits to the account behind the key. Per Firecrawl's billing docs: map 1 credit per call; search 2 credits per 10 results; scrape 1 credit per page. A free account has 1,000 credits a month, so a large run can use it up. docs.firecrawl.dev

Security advisories

No published security advisories as of Nov 2, 2026.

Technical details

Turning on read-only	No read-only mode is documented in SKILL.md. Use version control for files it writes.
Talks to	Firecrawl hosted service. The documentation sites and web pages you point it at
Sign-in and keys	SKILL.md declares FIRECRAWL_API_KEY as a required input. It does not say where the key is kept.
Scopes	Runs Firecrawl map, search and scrape commands with your API key Writes markdown, chunk, manifest and JSONL files under .firecrawl/ Spends credits on the Firecrawl account that owns your API key

Questions

What does the firecrawl-knowledge-base skill do?

It makes your agent build a knowledge base from web content. It scrapes a docs site or topic into markdown and organizes it as reference docs, RAG chunks, training data or a docs mirror.

How do I install it in Claude Code or Codex?

Use the install box at the top of this page. Each client has a plugin route and skills CLI commands. Install the Firecrawl CLI first (npm install -g firecrawl-cli, per its docs), then run firecrawl login or set FIRECRAWL_API_KEY.

Who pays for it?

You do, through the Firecrawl account that owns your API key. Firecrawl bills in credits, and a new account starts on the free plan with 1,000 credits a month and no card. Per its billing docs: map 1 credit per call; search 2 credits per 10 results; scrape 1 credit per page. Check your balance with firecrawl credit-usage.

Where are the files saved?

Under .firecrawl/<hostname>/<path>/index.md in your project, following the Firecrawl download-style convention named in SKILL.md. The rerun inputs list .firecrawl/ as the default output_dir.

firecrawl-knowledge-base vs firecrawl-knowledge-ingest?

Use knowledge-base for LLM-ready corpora built with map, search and scrape. SKILL.md for knowledge-ingest is aimed at portals that need browser navigation, sign-in, pagination or JavaScript rendering.

Is it safe to install?

It is a single SKILL.md file and bundles no scripts. When it runs, your agent calls Firecrawl with your API key and writes many files under .firecrawl/. Check the output folder and the depth setting (quick, thorough or exhaustive) before a large crawl.

Information

updated Nov 2, 2026

Repository: firecrawl/firecrawl-workflows
Folder: skills/firecrawl-knowledge-base

firecrawl/firecrawl-workflows skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 8

firecrawl-knowledge-base SKILL.md · github.com
firecrawl-workflows README · github.com
firecrawl-workflows plugin.json · github.com
firecrawl-workflows marketplace.json · github.com
firecrawl-workflows Codex plugin.json · github.com
Firecrawl CLI docs · docs.firecrawl.dev
Firecrawl billing and credits · docs.firecrawl.dev
firecrawl/cli README · github.com