the-time-masheen（马辛时间）

THE_TIME_MASHEEN

*Step in. Go back. Scrape the dead. Automate the living.*

Three-layer web intelligence stack. Pick the right layer — or combine all three.

Decision Tree

Need web data?
│
├─ Historical / "what did it look like before"?
│    └─ Wayback CDX API → scrape snapshot via Scrapling or web_fetch
│
├─ Need to click / log in / fill forms first?
│    └─ playwright-cli → authenticate → hand off to Scrapling
│
└─ Just current content?
     ├─ Static / simple       → scrapling get
     ├─ JS-heavy / React      → scrapling fetch --network-idle
     └─ Heavily protected sites → scrapling stealthy-fetch --solve-cloudflare

Layer 1 — Live Scraping (Scrapling)

Escalation path (always start at the top)

# 1. Static sites, blogs, docs
scrapling extract get "https://example.com" output.md

# 2. JS-heavy / React / Next.js / dynamic content
scrapling extract fetch "https://example.com" output.md --network-idle --wait 3000

# 3. Cloudflare / rendering-protected
scrapling extract stealthy-fetch "https://example.com" output.md --solve-cloudflare

Extract specific sections (saves tokens)

scrapling extract fetch "https://example.com" output.md --css-selector "main article"
scrapling extract get "https://example.com" output.md --css-selector ".pricing-table"

Rules:

• Always clean up temp files after reading
• Use .md output for readable text, .html only for structure parsing
• Use --css-selector to avoid giant HTML blobs

See references/scrapling.md for full CLI flags, spider framework, and Python API.

Layer 2 — Time Travel (Wayback Machine)

Find all snapshots of a URL

curl -s "https://web.archive.org/cdx/search/cdx?url=example.com&output=json&fl=timestamp,statuscode&filter=statuscode:200&limit=20"

One snapshot per year (change tracking)

curl -s "https://web.archive.org/cdx/search/cdx?url=example.com&output=json&collapse=timestamp:4&fl=timestamp,statuscode&filter=statuscode:200"

Scrape a specific point in time

# Scrapling for clean extraction:
scrapling extract get "https://web.archive.org/web/20230601000000/https://example.com/" archive.md

# Or read via web_fetch:
# web_fetch: https://web.archive.org/web/20230601000000/https://example.com/

Check if a URL has ever been archived

curl -s "https://archive.org/wayback/available?url=example.com" | python3 -m json.tool

See references/wayback.md for full CDX API reference and ia CLI usage.

Layer 3 — Interactive Automation (playwright-cli)

Use when the page requires login, clicking, or dynamic interaction before content is accessible.

# Open browser
playwright-cli open https://app.example.com

# Snapshot to get element refs
playwright-cli snapshot

# Interact
playwright-cli click e12
playwright-cli fill e5 "username@example.com"
playwright-cli press Tab
playwright-cli fill e6 "password"
playwright-cli press Enter

# Capture state
playwright-cli screenshot
playwright-cli eval "document.title"

# Close
playwright-cli close

Handoff pattern — authenticate then bulk scrape

# 1. playwright-cli open → log in → navigate to target
# 2. playwright-cli screenshot  # verify you're authenticated
# 3. scrapling extract get <url> output.md  # scrape while session is active

Combining All Three

Live vs. archived comparison (price changes, content drift, competitive intel)

# 1. Scrape current state
scrapling extract get "https://competitor.com/pricing" current.md

# 2. Find yearly snapshots
curl -s "https://web.archive.org/cdx/search/cdx?url=competitor.com/pricing&output=json&collapse=timestamp:4&fl=timestamp&filter=statuscode:200"

# 3. Scrape archived version from any year
scrapling extract get "https://web.archive.org/web/20230101000000/https://competitor.com/pricing" archive.md

# 4. Diff
diff archive.md current.md

Login-gated site — full extraction

# playwright handles auth → Scrapling does the bulk lift
playwright-cli open https://example.com/login
playwright-cli fill e5 "your@email.com"
playwright-cli fill e6 "password"
playwright-cli press Enter
playwright-cli screenshot  # verify you're in
scrapling extract get "https://example.com/members/content" output.md

Security

This skill opens real browser sessions and can scrape login-protected pages. A few things to understand before using it:

• You control the browser. playwright-cli drives a browser on your machine. It navigates to URLs you specify and interacts with elements you tell it to.
• All data stays local. Any session state used during automation exists only on your machine and is used only for the scraping task you initiate.
• Use only on sites you have access to. This skill is designed for legitimate web research — competitive intelligence, content monitoring, archival work, and accessing sites you have an account on. It is not a tool for unauthorized access.
• Review commands before running. As with any automation tool, understand what you're running before you run it.

CLI-Anything — make any software agent-native

When you need a full CLI harness for any desktop or web application:

# Install once in Claude Code
/plugin marketplace add HKUDS/CLI-Anything
/plugin install cli-anything

# Build a complete CLI for any software (7-phase pipeline)
/cli-anything:cli-anything ./target-app

# Iteratively refine
/cli-anything:refine ./target-app "focus on data export workflows"