Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/hex/claude-crawl/webnpx skills add hex/claude-crawl --skill webgit clone --depth 1 https://github.com/hex/claude-crawlWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00150 | $0.02107 |
| Opus 5 | $0.00075 | $0.01053 |
| Sonnet 5 | $0.00030 | $0.00421 |
| Haiku 4.5 | $0.00015 | $0.00211 |
Grade A, and why
web scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 173 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Web Content Operations
All web operations use shell scripts in ${CLAUDE_PLUGIN_ROOT}/scripts/. Output goes to
.claude-crawl/ in the project root to protect the context window from large content.
Before any operation: mkdir -p .claude-crawl/{reads,searches,crawls,extractions,reranks,ground,screenshots,pdfs}
WebFetch calls are automatically intercepted and routed through Jina Reader for cleaner
markdown output. Set CLAUDE_CRAWL_NO_INTERCEPT=1 to disable this behavior and use
WebFetch directly. Run check-auth.sh --verbose to see which services are available.
Routing Table
Choose the right tool based on what you have and what you need:
HAVE NEED USE
─────────────────────────────────────────────────────────────────────
Topic/question Search results jina-search.sh or fc-search.sh
URL, static page Clean markdown jina-read.sh
URL, JS-heavy/SPA Rendered content fc-scrape.sh
URL, need visual capture Screenshot or PDF cf-page.sh --format screenshot|pdf
URL, need structured data JSON extraction (AI) cf-page.sh --format json --ai-prompt "..."
Domain All URLs on site fc-map.sh
Multiple URLs, bulk Multi-page content cf-crawl-start.sh or fc-crawl.sh
Statement/claim Fact verification jina-ground.sh
Multiple results Ranked by relevance jina-rerank.sh
How to decide between similar tools
Search: Use jina-search.sh for general web search (fast, returns full content). Use
fc-search.sh when you need Firecrawl-specific features like --scrape (auto-scrapes each
result) or want an alternative search provider.
Single page fetch: Start with jina-read.sh — it's fastest and cheapest. If the result
is thin (< 100 words, no headings — the script warns you), escalate to fc-scrape.sh which
renders JavaScript. If you need Cloudflare's AI extraction or screenshots, use cf-page.sh.
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 173 lines · 150 tokens per session scan A 85776aab3f3b
web is a skill published in the GitHub repository hex/claude-crawl (1 stars, last pushed 5mo ago), licensed MIT. It adds 150 tokens to every session and 2,107 once invoked, about $0.0007 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
workspace-digest
Summarize the files saved in this assistant's shared workspace. Use when the user asks what is in their workspace, for a file inventory, or a digest of saved work.
workers-best-practices
Reviews and authors Cloudflare Workers code against production best practices. Load when writing new Workers, reviewing Worker code, configuring wrangler.jsonc, or checking for common Workers anti-patterns (streaming, floating promises, global state, secrets, bindings, observability). Biases towards retrieval from…
workers-code-review
Reviews Workers and Cloudflare Developer Platform code for type correctness, API usage, and configuration validity. Load when reviewing TypeScript/JavaScript using Workers APIs, wrangler.jsonc/toml config, or Cloudflare bindings (KV, R2, D1, Durable Objects, Queues, Vectorize, AI, Hyperdrive).
contributing
Use when contributing to the Cloudflare Docs repository — writing or editing documentation pages, choosing content types or components, adding changelog entries, reviewing docs, or learning how to contribute.
code-review
Review the changed lines of a single file in a pull request for bugs, correctness, error handling, security, and maintainability, and return structured findings.
conventions-check
Review a pull request's title, description, and scope against the repository's PR conventions.