PD2 Market Sniper

Automated trading assistant for Project Diablo 2 marketplace (softcore ladder).
Account: d1ngl3 on projectdiablo2.com

Prerequisites

Chrome must be running with remote debugging and logged into the d1ngl3 PD2 account:

Stop-Process -Name chrome -Force -ErrorAction SilentlyContinue
Start-Process "C:\Program Files\Google\Chrome\Application\chrome.exe" -ArgumentList "--remote-debugging-port=9222","--user-data-dir=C:\Users\jding\AppData\Local\Google\Chrome\User Data"

Main Workflows

1) Safe Scan Cycle

python scripts/sniper.py scan --max-price 0.5 --filters-per-cycle 12 --daily-limit 200 --max-pages 1

What it does:

• Rotates through only part of the saved filter list each cycle (default 12 filters)
• Enforces conservative pacing with 2.0-3.5s randomized delays between filters
• Supports bounded pagination with --max-pages (default 1 page)
• Adds randomized delay between page turns during multi-page scans
• Caps daily checks (default 200) via sniper_state.json
• Intercepts market XHR/fetch JSON when available, with DOM fallback if the API shape changes
• Expires old seen items after 36h so relists can surface again
• Auto-refreshes pd2.tools economy data daily
• Produces structured scan_results.json
• Writes assets/dashboard.html

2) Scan a Specific Filter

python scripts/sniper.py scan --filter-id 69e0c58f9fc33c4bc7fe0483 --max-price 0.5 --max-pages 3

3) Alert-Then-Offer Flow (PRIMARY WORKFLOW FOR CHAT)

CRITICAL: The agent IS the notification layer. The CLI only prints to stdout — it does NOT message the user. The agent must read the JSON output, format deals for the chat platform, and send them to the user.

Step-by-step:

1. Run operator-scan — captures JSON with ranked deals to scan_results.json.
2. Parse the JSON output — read the structured result from the CLI.
3. Format deals for chat — for each deal, send a concise card with:
   • Item name, price (HR), seller, score
   • Key stats/corruptions (if any)
   • Direct link
   • Screenshot image (attach via MEDIA:<path> if available)
   • Economy value comparison (if available)
4. Message the user — send each deal as a separate message to the chat channel (WhatsApp/Discord/etc).
5. Ask what they want to offer — explicitly prompt: "Reply with an HR amount (e.g. 0.3) to make an offer, or skip."
6. Wait for user reply — when the user sends a number, run reply-offer <amount>.
7. Confirm the result — tell the user if the offer went through or failed.

WhatsApp formatting (no markdown tables, no headers):

🎯 DEAL #1 (Score: 12.4)
⭐ Griffon's Eye
💰 Asking: 0.4 HR | 💎 Econ: 1.2 HR (66% off)
👤 Seller: someguy
🔗 https://www.projectdiablo2.com/market/listing/abc123

Stats:
- +1 Lightning Skills
- 20% FCR
- +15% Light Damage

Reply with an HR amount (e.g. 0.3) to offer, or skip.

Operator scan:

python scripts/sniper.py operator-scan --max-price 0.5 --filters-per-cycle 12 --daily-limit 200 --max-pages 1 --top 5

This stores the top result as the current pending confirmation target and writes structured JSON to scan_results.json.

Show the pending deal again:

python scripts/sniper.py pending

Submit a user reply amount against the pending deal:

python scripts/sniper.py reply-offer 0.3

Direct offer by URL:

python scripts/sniper.py offer --listing-url "https://www.projectdiablo2.com/market/listing/abc123..." --amount 0.3

One-click confirm using the last scan result:

python scripts/sniper.py confirm --deal-index 0 --amount 0.3

The sniper does not auto-price. It alerts, ranks, screenshots, and waits for the user to choose the number. The agent is responsible for delivering these alerts to the user via the active chat channel.

Other Commands

List saved filters:

python scripts/sniper.py filters

Show offer history stats:

python scripts/sniper.py history

Show the current pending confirmation deal as JSON:

python scripts/sniper.py pending --json

Refresh economy cache:

python scripts/sniper.py economy --force

Regenerate dashboard from saved data:

python scripts/sniper.py dashboard

Files and Data

• scripts/sniper.py - CLI entry point
• scripts/scanner.py - scan orchestration, pacing, API interception, deal collection
• scripts/offers.py - offer submission flow
• scripts/alerts.py - scoring and alert formatting
• scripts/economy.py - pd2.tools refresh and lookup
• scripts/history.py - seen item expiry, offer history, scan state, filter health
• scripts/dashboard.py - HTML dashboard generation
• scripts/parsers.py - API/DOM listing parsers
• scripts/config.py - shared config and filter list
• seen_items.json - dedup tracker with timestamps
• offer_history.json - offer log and status history
• sniper_state.json - cadence counters, rotation state, filter health, recent scans
• scan_results.json - latest structured scan output
• assets/all_economy.json - cached economy values
• assets/dashboard.html - scan dashboard
• screenshots/ - deal and offer screenshots

Scan Result Shape

Each scan returns structured JSON with:

• summary: timing, filter count, daily count, deal count
• filters: per-filter results, API endpoints hit, errors if any
• deals: ranked deals with score, direct link, screenshot path, seller, price, and economy context
• review: listings that need manual review because no HR price was parsed
• state: recent scan history and filter-health data

Filter Health

The sniper tracks whether filters return listings over time:

• Filters with no hits for 7+ days are marked slow
• Slow filters are still kept in rotation
• Nothing is auto-removed

Safe Usage Notes

• Conservative cadence is deliberate to reduce ban risk
• Keep cycles limited instead of hammering all 59 filters at once
• Use --max-pages 1 for normal sniper behavior, increase pages only for deliberate deep scans
• Avoid spamming the same listing, use offer_history.json as a sanity check
• Season 13 ladder resets April 24, 2026

Saved Filters

All existing saved filter IDs are preserved in scripts/config.py.