Every morning you should know: what’s moving in your categories, which competitors changed prices, what new products entered your space. Most sellers check none of this — because manually compiling it takes hours.

The irony is clear. Amazon sellers make decisions worth thousands of dollars every week, but the information that should ground those decisions sits scattered across browser tabs, platform dashboards, and half-finished spreadsheets. The data exists. The synthesis does not.

This article describes how to build a daily market dashboard that compiles itself — using one MCP command, one open-source server, and an AI agent that does the heavy lifting while you read your first coffee.

What a Daily Dashboard Should Cover

A useful morning brief for an Amazon seller answers a handful of questions:

Each of these questions can be answered with existing marketplace data. The bottleneck is not access. It is the act of checking.

To answer all five questions manually, a seller opens at least three tools, runs six to ten separate queries, cross-references numbers across tabs, and manually formats the findings into something readable. Done thoroughly, this is a 45-minute ritual. Done quickly, the corners cut are the ones that matter most — the outlier, the change, the signal that something shifted.

Why Sellers Skip It

The manual process fails for two reasons.

First, it does not scale. A seller tracking three categories and fifteen competitor ASINs has more data points to check than minutes in a morning. The ritual breaks down under its own weight. By Wednesday, the spreadsheet has gaps. By Friday, the habit has collapsed.

Second, it is fragile. A manual morning routine depends on the seller remembering to do it, having the time to do it, and not being pulled into an urgent operational issue — a shipment delay, a listing suppression, a customer escalation — that pushes market monitoring off the day entirely.

The result is a common pattern: sellers operate on stale information most of the time, supplemented by reactive deep-dives when something goes wrong. The daily pulse that would surface problems early and opportunities first never materializes.

This is not a discipline problem. It is a tooling problem. The process is designed for a human to execute, and humans are bad at repetitive multi-source data synthesis. Machines are good at it.

One MCP Server, One Prompt, One Morning Report

The sorftime-seller-agent is an open-source MCP (Model Context Protocol) server that exposes marketplace intelligence as tools an AI agent can call directly. It covers ASIN analysis, keyword research, category scouting, price tracking, review intelligence, and profitability estimation — across multiple Amazon marketplaces.

The value proposition is not new data. It is a new access pattern. Instead of the seller going to the data — opening dashboards, running queries, copying numbers — the data comes to the seller, compiled and prioritized, in a single structured report.

Here is the setup:

git clone https://github.com/DannylydST/sorftime-seller-agent
cd sorftime-seller-agent
python3 scripts/install.py

Example output: top 100 category report showing brand concentration and review distribution. Example output: top 100 category report showing brand concentration and review distribution.

Once installed and connected to an MCP-compatible AI agent (Claude Code, Codex, Cursor, and others all support MCP), a single prompt generates the daily dashboard:

Run a daily market dashboard for my Amazon business. Here are the parameters:

CATEGORIES TO WATCH:
- "stainless steel water bottle" (Amazon US)
- "camping cookware" (Amazon US)

COMPETITOR ASINS TO TRACK:
- B0XXXXXXXXX1 (my main competitor's flagship)
- B0XXXXXXXXX2 (new entrant from last month)
- B0XXXXXXXXX3 (price leader in the category)

For each category:
1. Show new ASINs that entered the top 20 in the last 24 hours (if available)
   or top 20 movements over the last 7 days. Include BSR, price, review count,
   and estimated monthly sales for each.
2. Flag any product in the top 20 that moved more than 5 positions up or down.
3. Report the average and median price for the top 20 and note any listing
   priced more than 20 percent below the median.

For each tracked ASIN:
4. Compare current price to price 7 days ago. Flag if changed by more than 5 percent.
5. Check current review count against last check. Flag any burst of 10+ reviews
   in a week — possible promotional activity.
6. Show organic keyword rankings for the top 5 keywords, with position changes.
7. Report current BSR and BSR trend direction (rising, falling, stable).

Across all tracked data:
8. Highlight the 3 most actionable findings at the top of the report.
9. Include a "since yesterday" section if previous data is available.

Format the output as a structured morning brief with clear sections, numbers,
and one-line verdicts for each finding.

The AI agent calls the necessary MCP tools — product search, price history, keyword rankings, category reports, review data — in sequence, applies the filtering logic, and returns a single structured report. The seller reads it. The seller decides. The agent handled the reading.

What the Output Looks Like

A typical morning report runs somewhere between 400 and 800 words depending on how many categories and ASINs are tracked. The structure mirrors the prompt:

  1. Top of report: three most actionable findings. These are the signals that warrant a decision today — a competitor dropped price by 12 percent, a new ASIN entered the top 10 with zero reviews suggesting a launch play, a keyword your listing ranks for just tripled in search volume.

  2. Category snapshots. For each category tracked, a compact table showing the top 20 with flags for movement. New entrants bolded. Price outliers marked. The seller scans for anomalies, not for completeness.

  3. Competitor watch. Per-ASIN tables with current price versus prior, review count delta, keyword position changes, and BSR trend. The agent flags the deltas — the seller does not hunt for them.

  4. Keyword radar. A short section listing terms with significant search volume movement in the tracked categories. Seasonal spikes, trending terms, declining terms — whatever moved since the last check.

The report is designed to be read in under three minutes. If nothing changed, the report says so and the seller moves on. The value is not in every report being dramatic. It is in the ones that are dramatic being seen.

Customizing for Your Business

The prompt above is a template. Every seller’s dashboard should reflect their specific competitive landscape. Common customizations include:

The prompt is the config file. Changing what the dashboard covers is a matter of editing the English description — not adjusting a dashboard widget, not reconfiguring an integration, not writing a script.

What This Approach Does Not Do

It does not replace strategic thinking. The agent identifies signals. The seller interprets them. A competitor dropping price by 10 percent is a data point. Whether to match, ignore, or reposition is a business decision that depends on margins, brand positioning, inventory levels, and long-term goals — factors the agent cannot weigh.

It does not guarantee the data is real-time to the second. Marketplace APIs have refresh intervals. BSR updates with some lag. Review counts may be cached. The dashboard is a morning pulse, not a ticker. For intra-day monitoring of fast-moving events — lightning deals, flash sales, viral surges — supplemental real-time checks are appropriate.

It requires an MCP-compatible AI agent. The sorftime-seller-agent is not a standalone web dashboard. It is a tool server that plugs into an AI agent. If your workflow does not already involve an AI coding or research assistant, there is a one-time setup step: install Claude Code (free tier), connect the MCP server, and start asking questions.

It requires an API key. A free key is available at open-intl.sorftime.com. The free tier covers the query volume needed for a daily dashboard across a reasonable number of categories and ASINs. Higher-volume commercial use requires a paid plan.

The Shift

The difference between a daily dashboard built with MCP and a daily dashboard built manually is not speed — though the MCP version is faster. It is reliability. A manual morning routine depends on the seller showing up and executing it. An MCP-driven routine depends on a configuration file and a schedule. One of those is consistent. The other is human.

For an Amazon seller running a multi-product, multi-category business, dependable market intelligence is not a luxury. It is the difference between reacting to problems after they have cost money and seeing them before they do.

Get the dashboard running.

Register for a free API key at open-intl.sorftime.com, then run:

git clone https://github.com/DannylydST/sorftime-seller-agent
cd sorftime-seller-agent
python3 scripts/install.py

Connect the server to your MCP-compatible AI agent. Copy the dashboard prompt from above into a scheduled task set to run each weekday morning. The report will be waiting.

sorftime-seller-agent on GitHub Free API Key

References

[1] Sorftime Seller Agent GitHub Repository — https://github.com/DannylydST/sorftime-seller-agent [2] Sorftime International MCP Portal — https://open-intl.sorftime.com [3] Model Context Protocol Specification — https://modelcontextprotocol.io —