Automate CS2 Skin Trading Bots with the CSBoard API

The CSBoard API is designed for automation. Every endpoint is deterministic, cursor-paginated, and rate-limit-aware, which makes it straightforward to build price monitors, sniping bots, arbitrage tools, and full market data pipelines. This guide covers the key patterns you need to build reliable automated systems.

AI Agent Integration

If you are building an AI agent or LLM-powered tool, the fastest way to give it access to the full API surface is via the machine-readable spec at /llms.txt. That file contains the complete API in a single, compact document — no parsing required. For agents that support the Model Context Protocol (MCP), add the CSBoard server to your MCP config:

{
  "mcpServers": {
    "csboard": {
      "url": "https://csboard.com/mcp",
      "headers": {
        "Authorization": "Bearer csb_pub_..."
      }
    }
  }
}

Once connected, the agent can call any CSBoard endpoint as a tool — browsing listings, checking prices, and placing orders — without you writing any glue code.

Polling for New Listings

To detect new listings as they appear without re-scanning the entire catalog, use sort=newest combined with cursor pagination. On each poll cycle, walk pages until you reach a listing ID you have already seen, then stop.

# First poll — fetch the newest 50 listings and record next_cursor
curl "https://csboard.com/v1/listings?sort=newest&limit=50" \
  -H "Authorization: Bearer csb_pub_..."

import time
import httpx

API_KEY = "csb_pub_..."
BASE_URL = "https://csboard.com/v1"
SEEN_IDS: set[str] = set()

def poll_new_listings(category: str = "Knife") -> list[dict]:
    """Return all listings added since the last poll."""
    new_items = []
    cursor = None

    while True:
        params = {"sort": "newest", "limit": 50, "category": category}
        if cursor:
            params["cursor"] = cursor

        resp = httpx.get(
            f"{BASE_URL}/listings",
            params=params,
            headers={"Authorization": f"Bearer {API_KEY}"},
        )
        resp.raise_for_status()
        data = resp.json()

        for item in data["items"]:
            if item["id"] in SEEN_IDS:
                # Reached a listing we already processed — stop paging
                return new_items
            SEEN_IDS.add(item["id"])
            new_items.append(item)

        if data["next_cursor"] is None:
            break
        cursor = data["next_cursor"]

    return new_items

while True:
    fresh = poll_new_listings(category="Knife")
    if fresh:
        print(f"Found {len(fresh)} new listing(s)")
    time.sleep(30)  # Respect rate limits — poll every 30 seconds

Seed SEEN_IDS on startup by running one full poll without acting on the results. That way, you only trigger actions on listings that appear after your bot starts — not on everything that was already live.

Price Monitoring Bot

Poll GET /v1/prices on a schedule, compare each item’s min_price_usd against your target threshold, and trigger an alert or automated buy when the price drops below it.

import httpx
import time

API_KEY = "csb_pub_..."
BASE_URL = "https://csboard.com/v1"

# Items to watch: market_hash_name → max price willing to pay
WATCHLIST = {
    "AK-47 | Redline (Field-Tested)": 12.00,
    "AWP | Asiimov (Field-Tested)": 55.00,
    "Karambit | Fade (Factory New)": 800.00,
}

def check_prices() -> list[dict]:
    """Return watchlist items whose price has dropped to or below threshold."""
    alerts = []

    for name, threshold in WATCHLIST.items():
        resp = httpx.get(
            f"{BASE_URL}/prices",
            params={"search": name, "max_price": threshold},
            headers={"Authorization": f"Bearer {API_KEY}"},
        )
        resp.raise_for_status()
        data = resp.json()

        for row in data["items"]:
            if row["market_hash_name"] == name and row["min_price_usd"] <= threshold:
                alerts.append({
                    "name": name,
                    "price": row["min_price_usd"],
                    "qty": row["qty"],
                    "threshold": threshold,
                })

    return alerts

while True:
    try:
        hits = check_prices()
        for alert in hits:
            print(
                f"ALERT: {alert['name']} @ ${alert['price']:.2f} "
                f"(threshold ${alert['threshold']:.2f}, qty {alert['qty']})"
            )
            # → trigger_buy(alert["name"]) or send_notification(alert)
    except httpx.HTTPStatusError as e:
        if e.response.status_code == 429:
            retry_after = int(e.response.headers.get("Retry-After", 60))
            print(f"Rate limited — waiting {retry_after}s")
            time.sleep(retry_after)
            continue
        raise

    time.sleep(60)  # Poll once per minute

min_price_usd from /v1/prices is an indicative grouped snapshot — it can lag the live per-item price. Always fetch the specific listing from /v1/listings and use its price_usd as the authoritative price before placing an order.

Safe Buying Automation

Automated buying requires more defensive coding than manual purchasing. Follow these four rules unconditionally. 1. Always send max_price_usd Your ceiling is enforced atomically inside the balance debit. Without it, a price spike between your listing fetch and your order execution can result in an unexpected charge. Set max_price_usd to the price_usd you observed, plus a small buffer if you are willing to absorb minor slippage:

listing_price = 14.37
max_price = round(listing_price * 1.02, 2)  # Allow up to 2% slippage

2. Always use an Idempotency-Key Generate one UUID v4 per order attempt. If your request times out or returns a network error, retry with the same key — the server replays the original result instead of executing a second purchase.

import uuid
import httpx

def place_order(item_ids: list[str], max_price: float) -> dict:
    idempotency_key = str(uuid.uuid4())  # One key per attempt

    resp = httpx.post(
        "https://csboard.com/v1/orders",
        headers={
            "Authorization": f"Bearer {API_KEY}",
            "Idempotency-Key": idempotency_key,
        },
        json={"item_ids": item_ids, "max_price_usd": max_price},
    )
    resp.raise_for_status()
    return resp.json()

3. Handle 429 with Retry-After Rate-limited responses include a Retry-After header (seconds). Always read that value and sleep for exactly that duration — do not use a fixed backoff, as it may be shorter or longer than required. 4. Handle 400 price_moved gracefully A 400 price_moved means the price moved past your ceiling — nothing was charged. Decide whether to re-fetch the listing, update max_price_usd, and retry, or abandon the opportunity:

def buy_with_retry(item_id: str, max_price: float, retries: int = 3) -> dict | None:
    for attempt in range(retries):
        try:
            return place_order([item_id], max_price)
        except httpx.HTTPStatusError as e:
            if e.response.status_code == 400 and e.response.json().get("code") == "price_moved":
                data = e.response.json()
                current = data.get("current_total_usd", max_price)
                print(f"Price moved to ${current:.2f}, was ${max_price:.2f}")
                # Decide: abort, or raise ceiling and retry
                if current <= max_price * 1.05:  # Accept up to 5% over
                    max_price = current
                    continue
                print("Price spike too large — aborting")
                return None
            if e.response.status_code == 429:
                retry_after = int(e.response.headers.get("Retry-After", 60))
                time.sleep(retry_after)
                continue
            raise
    return None

Bulk Data Pipeline

For comparison sites, analytics dashboards, or any system that needs the full catalog, use the snapshot endpoint instead of paginating through /v1/prices.

# Download the full snapshot
curl -s "https://csboard.com/v1/prices/snapshot.ndjson.gz" \
  -H "Authorization: Bearer csb_pub_..." \
  -o snapshot.ndjson.gz

gunzip -c snapshot.ndjson.gz | wc -l   # Count total rows

Use ETags to avoid re-downloading an unchanged snapshot. Store the ETag header value from each 200 response and send it back as If-None-Match on the next request. A 304 Not Modified response means your local copy is still current.

import gzip
import json
import httpx

SNAPSHOT_URL = "https://csboard.com/v1/prices/snapshot.ndjson.gz"
stored_etag: str | None = None

def refresh_snapshot() -> list[dict] | None:
    """Download snapshot only if it changed. Returns rows, or None if unchanged."""
    global stored_etag

    headers = {"Authorization": f"Bearer {API_KEY}"}
    if stored_etag:
        headers["If-None-Match"] = stored_etag

    resp = httpx.get(SNAPSHOT_URL, headers=headers)

    if resp.status_code == 304:
        print("Snapshot unchanged — skipping")
        return None

    resp.raise_for_status()
    stored_etag = resp.headers.get("ETag")

    rows = []
    for line in gzip.decompress(resp.content).splitlines():
        if line.strip():
            rows.append(json.loads(line))

    print(f"Loaded {len(rows)} price rows")
    return rows

The snapshot endpoint is rate-limited to 1 request per minute. Structure your pipeline to use it as a base layer refreshed every few minutes, and overlay real-time updates from /v1/prices for items you are actively monitoring.

Best practices checklist

Before going to production, verify all of the following:

✅ You send max_price_usd on every POST /v1/orders call
✅ You generate a fresh UUID v4 Idempotency-Key for every order attempt
✅ Your retry logic reuses the same Idempotency-Key on network-error retries
✅ Your 429 handler reads and sleeps for Retry-After, not a hardcoded value
✅ Your 400 price_moved handler decides explicitly whether to retry or abort
✅ You read price_usd from /v1/listings (not /v1/prices) before placing an order
✅ You use ETags with the snapshot endpoint to avoid redundant downloads
✅ Your API key has trading enabled only if your bot is intended to buy

​AI Agent Integration

​Polling for New Listings

​Price Monitoring Bot

​Safe Buying Automation

​Bulk Data Pipeline

​Best practices checklist

​Related pages

AI Agent Integration

Polling for New Listings

Price Monitoring Bot

Safe Buying Automation

Bulk Data Pipeline

Best practices checklist

Related pages