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/agentcomputerai/torch/amazonnpx skills add AgentComputerAI/torch --skill amazongit clone --depth 1 https://github.com/AgentComputerAI/torchWhat 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 | $0.00093 | $0.02368 |
| Opus 5 | $0.00046 | $0.01184 |
| Sonnet 5 | $0.00019 | $0.00474 |
| Haiku 4.5 | $0.00009 | $0.00237 |
Grade A, and why
amazon scanned grade A with 1 finding 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
description: Proven scraping playbook for amazon.com search result pages (/s?k=...). CloudFront + CAPTCHA wall blocks bare curl with HTTP 503, but a real Chrome session via the real Chrome debug port walks right through How it starts
The opening of the file, as written. The whole thing — 158 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Amazon (amazon.com)
Amazon search pages are server-rendered HTML behind CloudFront. The only real obstacle is IP/UA reputation on anonymous clients (curl →
HTTP/2 503+x-cache: Error from cloudfront). Attaching to the user's real Chrome profile via127.0.0.1:9222defeats the bot wall on the first request — no stealth plugin, no proxy, no captcha solver required. From there it's pure cheerio on 22s-search-resultcards per page.
Detection
| Signal | Value |
|---|---|
| CDN | CloudFront (via: 1.1 ...cloudfront.net, x-amz-cf-id, x-amz-cf-pop) |
| Framework | Server-rendered HTML (no SPA, no hydration payload needed) |
| Anti-bot | CloudFront + Amazon bot wall. Bare curl with a fake UA → HTTP 503, x-cache: Error from cloudfront. Escalates to Robot Check / CAPTCHA page if you push harder from a clean IP. |
| Auth | Not required for /s search pages. |
| robots.txt | Disallows /s?, but Amazon doesn't enforce it in-band — just don't abuse it. |
Architecture
/s?k=<query>&page=<n>returns full HTML with 22 result cards per page, already populated (no client-side hydration needed for the fields we care about).- Each card is wrapped in
div[data-component-type="s-search-result"]withdata-asin="<ASIN>". - Title lives in
h2 span(text). Link ish2 a[href](relative, may be an/sspa/click?...url=...redirect for sponsored slots). - Price is
.a-price > .a-offscreen→"$83.59". - Rating and review count are both encoded in the aria-label of the review link:
"Rated 4.5 out of 5 stars by 1,234 reviews. Go to review section."— regex it out, don't trust the nested star spans (they render with inconsistent text). - Image:
img.s-image[src]. - Sponsored flag: presence of
.puis-sponsored-label-text/.s-sponsored-label-textinside the card.
Strategy used
- Phase 0 (curl) — failed.
curl -A <chrome UA> https://www.amazon.com/s?k=...→HTTP/2 503from CloudFront. Confirmed this is IP/UA reputation, not a missing header. - Phase 1 (framework) — skipped. No Next.js
__NEXT_DATA__, no clean JSON API on public search. The HTML is the API. - Phase 2 (browser) —
puppeteer.connect({ browserURL: the real Chrome debug port })to the user's real Chrome. First request loaded full results, no challenge, no captcha. Thenpage.content()→ cheerio. Do not use a fresh Chromium + stealth here; it's a wasted escalation step on Amazon.
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 · 158 lines · 93 tokens per session scan A 9f02d88f7402
amazon is a skill published in the GitHub repository AgentComputerAI/torch (5 stars, last pushed 4mo ago), licensed MIT. It adds 93 tokens to every session and 2,368 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
tiny-web-crawler
Crawl from one or more starting web pages, fetch readable content, search within pages, follow relevant links, and stop when the requested information is found or a bounded limit is reached.
playwright-cli
Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, test web applications, or extract information from web pages.
browserstack
../../../engineering-team/playwright-pro/skills/browserstack/SKILL.md.
analyze-performance-traces
Analyze Chrome, Chromium, Electron, React DevTools, or Perfetto-compatible JSON traces and audit user-reported profiling findings without loading large artifacts into context; prove trigger-to-render/layout chains, separate measured facts from source inference, find exact code choke points, classify forced layout and…
feature-demo-recording
Record a demo video of a web feature from a real browser. Two modes -- a NARRATED film where measured voiceover drives the timeline (designed slides, subtitles, punch-in camera, rendered from an HTML timeline), and a SILENT evidence clip for a PR or a QA pass. Use when the user asks to record a video, demo, or screen…
browser-recording
Record a browser flow as a video/GIF for evidence — animations, transitions, and multi-step interactions that a still screenshot cannot prove. Drives the project's own Playwright through a bundled runner, then converts to mp4 + GIF via ffmpeg. Use when the user asks to record a demo, capture a GIF or video of the UI…