Competiprice
← Back to the blog

Ask your AI agent about competitor prices

Compare offers, export price history and manage alerts in Claude, Codex or another AI agent using the Competiprice MCP connector.

Use case 3 min read

Imagine you sell a pair of headphones for €99. A competitor listed them at €105 yesterday and now charges €95. Before your morning trading meeting, ask your agent, “Which competitors dropped below my price since yesterday? Show me the previous price, the new price, when it was checked and the link.” The headphones would appear in this example because the competitor moved from above your price to below it.

The Competiprice MCP (Model Context Protocol) connector gives Claude, Codex and other compatible agents access to the observations recorded in your Competiprice account. The timestamp tells you when Competiprice last checked the offer. A newly added URL starts collecting observations on your account’s monitoring schedule.

Seven ways to look up monitoring data

What to ask forWhat your agent can retrieve
List monitored productsCatalog entries, references, product IDs and URL counts, including products that have no URLs yet.
Compare offers for a productFind your offer and competing offers by name, reference or EAN. See prices, availability, links and the time of the latest observation.
Find price changesRises, drops and offers that crossed below your price. Filter by date, domain or percentage change.
Get product historyChronological observations for one ID, a list of IDs or the entire catalog with the * wildcard, within your plan’s history window.
List alert rulesThe rules you have set up, the URLs they apply to and the thresholds that trigger them.
Review triggered alertsEvents recorded during the period, with the URL and the values observed when the alert fired.
Export statisticsA paginated CSV or JSON export covering the summary, catalog, per-product and per-URL statistics, observations, price changes, alert rules, alert events and daily usage. Detailed sections follow the read permissions you grant.

Six actions you can take from a conversation

After spotting a price drop, try “On this competitor URL, create an alert for the next price drop of at least 5%.” Granting the agent permission to create alerts lets it save that rule. Deleting rules has its own switch.

  • Products. Create a product with its name and reference, or delete a specific product and its monitoring data.
  • URLs. Add a competitor page to an existing product, or remove a URL along with its history. Removing the last URL keeps the product in your catalog.
  • Alerts. Create or delete a rule for a price drop, a price rise, a target price, a competitor crossing below your price or a stock change.
Six write permission switches. Adding monitored URLs and creating alerts are enabled.
An example for an agent that tracks new offers and sets alerts. All three deletion permissions are off.

Each agent has its own switches in Settings → AI agents. Write access requires your consent. For an older read-only connection, reconnect with the monitoring:manage scope. You can then turn off any individual action or revoke the whole connection. Deleting a product along with its URLs and their alerts also requires the corresponding deletion permissions.

How many reports can I run?

All agents on the account share one allowance. Each offer comparison, price change query, history query or export page counts as one report. One report can run at a time.

PlanMCP requests per minuteReports per rolling hourMinimum time between starts
Discovery302415 s
Standard906010 s
Comfort1801206 s
Premium2402404 s

For a first check, ask your agent to list five references from your catalog. You can see exactly which records it has access to. Follow the quick start for Claude, Codex or another MCP client.