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/aphrody-code/bxc/scraper-recipenpx skills add aphrody-code/bxc --skill scraper-recipegit clone --depth 1 https://github.com/aphrody-code/bxcWhat 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 | $0.00066 | $0.01096 |
| Opus 5 | $0.00033 | $0.00548 |
| Sonnet 5 | $0.00013 | $0.00219 |
| Haiku 4.5 | $0.00007 | $0.00110 |
Grade A, and why
scraper-recipe 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 2d ago.
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 — 103 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Bxc scraper recipe
Quand invoquée via /scraper-recipe, cette skill produit un fichier scraper complet à partir des inputs suivants :
- URL ou pattern d'URL (obligatoire)
- Schéma cible (description naturelle, ex : "titre + prix EUR + stock")
- Profil souhaité (optionnel, défaut =
static) - Mode (one-shot dans
examples/<name>.ts, ou sub-package danspackages/<name>/src/)
API réelle (0.4.0)
L'API publique est Browser.newPage() → page.* → page.close(). Il n'existe
pas de Browser.fetch() ni Browser.scrape(). La feature LLM-extract a été
supprimée — l'extraction se fait par sélecteurs CSS (page.$$) ou Markdown
(page.markdown()), puis Zod.parse().
import { Browser } from "@aphrody/bxc"; // singleton
import { googleSearchRich } from "@aphrody/bxc/google"; // si recherche
Surface page utile : goto(url,{timeoutMs}), content() (HTML), markdown()
(GFM, fallback JS si cdylib absente), $(sel)/$$(sel) (handles → .textContent()
/.getAttribute(name)), title(), screenshot(), evaluate(fn) (profils JS),
close().
Workflow
-
Lis le contexte :
src/api/browser.ts—Browser.newPage(opts)+ classePage(API ci-dessus)src/api/browser.ts(PageOptions) — profils valides :static | http | fast | stealth | maxsrc/storage/Dataset.ts— store JSONL interne (Datasetnon exporté du package : import relatif ou JSONL manuel)
-
Décide l'extraction :
- HTML structuré + sélecteurs CSS suffisent →
page.$$()+Zod.parse() - Contenu prose / page entière →
page.markdown()
- HTML structuré + sélecteurs CSS suffisent →
-
Génère le fichier avec ce template :
import { z } from "zod";
import { Browser } from "@aphrody/bxc";
const Schema = z.object({
// ... champs selon la demande utilisateur
title: z.string(),
});
type Item = z.infer<typeof Schema>;
async function scrape(url: string): Promise<Item> {
const page = await Browser.newPage({ profile: "static" }); // static|http|fast|stealth|max
try {
await page.goto(url, { timeoutMs: 30_000 });
const titleEl = await page.$("h1");
const raw = {
title: (await titleEl?.textContent()) ?? "",
// ... autres champs via page.$$/$ ou page.markdown()
};
return Schema.parse(raw);
} finally {
await page.close();
}
}
const urls = [/* ... */];
const out = Bun.file("dataset.jsonl").writer();
try {
for (const url of urls) {
try {
const item = await scrape(url);
out.write(JSON.stringify(item) + "\n");
console.log(`OK ${url}`);
} catch (err) {
console.error(`FAIL ${url}`, err);
}
}
} finally {
await out.end();
await Browser.close();
}
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.
- 2d ago First seen · 103 lines · 66 tokens per session scan A dc4b76d3636c
scraper-recipe is a skill published in the GitHub repository aphrody-code/bxc (2 stars, last pushed 2d ago), licensed Apache-2.0. It adds 66 tokens to every session and 1,096 once invoked, about $0.0003 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
obscura
Operate and validate Obscura for JavaScript page loading, stealth browsing, anti-fingerprinting, tracker blocking, screenshots and visual comparison, CDP automation with Puppeteer or Playwright, screencasting, PDF export, MCP browser interaction, and web extraction. Use when running Obscura against deterministic…
pinchtab
Use this skill when a task needs browser automation through PinchTab: open a website, inspect interactive elements, click through flows, fill out forms, scrape page text, reuse a dedicated automation profile with user approval, export screenshots or PDFs, manage multiple browser instances, or fall back to the HTTP API…
pinchtab-mcp
Use this skill when a task requires browser automation through PinchTab's MCP server connected to a remote browser instance. Covers navigation, element interaction, data extraction, form filling, multi-step flows, and session management via MCP tools.
pinchtab-stealth-score
Run the PinchTab stealth-score sweep against 15 bot-detection / fingerprint sites (sannysoft, rebrowser, deviceandbrowserinfo, iphey, whoer, browserscan, pixelscan, fingerprint-scan, incolumitas, fvision, amiunique, browserleaks, creepjs, coveryourtracks, fingerprint-demo). Starts a Docker PinchTab container per…
browser-use-terminal
Direct browser control via the Browser Use Terminal CLI. Use when the user wants to automate, scrape, test, or interact with web pages — you drive the browser yourself with Python helpers.
unicli
Comprehensive guide to Uni-CLI — the open Agent-Computer Interface runtime for real software. Trigger when the user needs to fetch data from websites (Twitter, Bilibili, HackerNews, GitHub, Reddit, Bloomberg, Zhihu, WeChat, and hundreds more); interact with news, finance, social, academic, shopping, or video…