Provides forecasting tools for Shopify stores, enabling revenue and demand forecasting, promotion analysis, anomaly detection, and scenario planning using historical order data.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@shopify-forecast-mcp What does next month look like?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
⚠️ v0.1.0 Alpha — Early release. API surface may change before v0.2. Feedback welcome: open an issue.
Merchant-native MCP server that connects Google's TimesFM 2.5 time-series foundation model to your Shopify store — so your AI assistant can answer "what does next month look like?" with a real forecast grounded in your order history.
No dashboards, no exports, no per-store training. Works with Claude Desktop, Claude Code, Cursor, and any MCP-compatible AI client.
Shopify has four official MCP servers, all buyer-facing or developer-facing. None serve merchant operations: forecasting, demand planning, promo analysis, anomaly detection.
Existing third-party tools either use weak models (moving averages, Prophet) or lock insights inside closed SaaS dashboards. This one:
Runs TimesFM 2.5 (Google's 200M-param foundation model) — state of the art on the GIFT-Eval retail benchmark
Pulls directly from Shopify Admin GraphQL with bulk operations, refund-aware normalization, multi-currency, and cost-based rate limiting
Returns markdown tables with confidence bands that render natively in your MCP client
Ships as a single uvx command — zero manual Python setup
Is MIT licensed, free forever
Related MCP server: Shopify MCP Server
Three steps to a working forecast in under 5 minutes:
uv (once)# macOS / Linux
curl -LsSf https://astral.sh/uv/install.sh | sh
# Windows (PowerShell)
irm https://astral.sh/uv/install.ps1 | iexFollow docs/SETUP.md to create a custom app, enable the required scopes (read_orders, read_all_orders, read_products, read_inventory), and generate an access token.
Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows) and add:
{
"mcpServers": {
"shopify-forecast": {
"command": "uvx",
"args": ["shopify-forecast-mcp"],
"env": {
"SHOPIFY_FORECAST_SHOP": "mystore.myshopify.com",
"SHOPIFY_FORECAST_ACCESS_TOKEN": "shpat_xxxxxxxxxxxxxxxxxxxxxxxx"
}
}
}
}Restart Claude Desktop, then ask: "What does next month look like?"
First run downloads TimesFM 2.5 weights (~400MB, one-time). Subsequent forecasts are <10 seconds.
Prefer the terminal first? You can verify your setup before wiring Claude Desktop by running
uvx shopify-forecast-mcpdirectly — it launches the MCP server on stdio and will exit cleanly on Ctrl-C when no client is attached. Or exercise the CLI:uvx --from shopify-forecast-mcp shopify-forecast revenue --horizon 7.
Alpha pre-release: if installing v0.1.0-rc1, use
"args": ["--prerelease=allow", "shopify-forecast-mcp@0.1.0rc1"]instead.
See also: docs/SETUP.md for Claude Code + generic MCP client setup, Docker install, and multi-store configuration.
Drop these into your AI client after setup:
Revenue forecasting — "What does next month look like?" Returns a daily-granularity revenue forecast for the next 30 days with an 80% confidence band (q10–q90).
Demand + reorder alerts — "Which SKUs need to be reordered in the next 2 weeks?" Returns top-N SKUs with projected demand vs current inventory, flagging stockout risk.
Promo analysis — "How did Black Friday perform vs last year?" Returns revenue lift, order lift, AOV change, discount depth, and post-promo hangover estimate for both windows side by side.
Scenario planning — "Compare 3 promo scenarios for December: 10% off, 20% off + free shipping, and BOGO." Returns 3 differentiated forecasts in one markdown response with per-scenario revenue, units, and margin implications.
Seven MCP tools, full reference in docs/TOOLS.md:
Tool | Purpose |
Store-level revenue forecast with confidence bands | |
Product/collection/SKU demand + reorder alerts | |
Past promo vs baseline — lift, AOV, hangover | |
Flag days outside forecast quantile bands | |
Year-over-year / month-over-month comparison | |
What-if forecasting across 2-4 scenarios | |
Explain learned seasonal patterns |
Two-layer design: a pure-Python core library (Shopify client, time-series shaping, TimesFM forecaster, analytics, covariates) wrapped by a thin MCP server and a matching CLI. Core is importable and testable without the MCP runtime.
Dual-backend Shopify access: DirectBackend (httpx + access token, used in Docker and when SHOPIFY_FORECAST_ACCESS_TOKEN is set) or CliBackend (shopify store execute — browser OAuth, no token required, host-only).
Full diagrams + design decisions in docs/ARCHITECTURE.md.
Minimum required env vars:
Variable | Purpose |
| Your store domain (e.g. |
| Admin API access token (from custom app) |
See docs/SETUP.md for the full env var table, multi-store config, Docker env passing, and optional tuning knobs.
A standalone shopify-forecast CLI wraps the same core library without the MCP runtime — useful for scripting, cron, and CI:
uvx --from shopify-forecast-mcp shopify-forecast revenue --horizon 30
uvx --from shopify-forecast-mcp shopify-forecast demand --group-by product --top-n 10
uvx --from shopify-forecast-mcp shopify-forecast promo --start 2025-11-24 --end 2025-11-30
uvx --from shopify-forecast-mcp shopify-forecast compare --period-a 2024-11:2024-12 --period-b 2025-11:2025-12Add --json to any verb for machine-readable output. See docs/SETUP.md#cli-usage for the full CLI reference.
Run the MCP server (or any CLI verb) without installing Python:
docker run --rm -i \
-e SHOPIFY_FORECAST_SHOP=mystore.myshopify.com \
-e SHOPIFY_FORECAST_ACCESS_TOKEN=shpat_xxx \
ghcr.io/mcostigliola321/shopify-forecast-mcp:latestTwo image variants:
:latest — lazy model download on first call (smaller image, ~1.5GB)
:bundled — TimesFM weights baked in (larger image, ~2.5GB, offline-capable)
Browser-based OAuth does NOT work in containers; Docker mode requires the access-token env var. See docs/SETUP.md#docker.
v0.1.0 is the first public alpha, covering the full MVP (7 MCP tools + 4 CLI verbs + dual-backend). Future: see .planning/ROADMAP.md.
Feedback and bug reports welcome at GitHub Issues. For code contributions, see docs/ARCHITECTURE.md for the two-layer design and open a draft PR early.
MIT. TimesFM 2.5 weights are Apache 2.0 (compatible).
Analyze a past promotion's impact vs baseline: revenue lift, order lift, AOV change, post-promo hangover, and product cannibalization.
| Name | Required | Description | Default |
|---|---|---|---|
| params | Yes |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description must convey behavioral traits. It lists outputs but does not describe side effects, data dependencies, or constraints. It is adequate but not comprehensive.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single well-structured sentence, front-loading the core purpose and listing key analytical components. Every word adds value without redundancy.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Despite having an output schema, the description lacks information about data requirements, time frame limitations, or how it differs from related tools. It covers the core analysis but misses important contextual details for an agent.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The schema (as shown) includes descriptions for each parameter, but context signals report 0% schema description coverage. The tool description adds nothing about parameters, which is insufficient given the low coverage.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool analyzes past promotion impact vs baseline, listing specific metrics like revenue lift and cannibalization. This distinguishes it from siblings like forecast_demand or compare_periods, which serve different purposes.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides no guidance on when to use this tool compared to alternatives such as compare_periods or detect_anomalies. No mention of prerequisites, context, or exclusions.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Compare two time periods across revenue, orders, units, AOV, discount rate, and units per order.
| Name | Required | Description | Default |
|---|---|---|---|
| params | Yes |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries full burden. It only states the purpose but no behavioral traits like read-only nature, required permissions, or side effects. For a comparison tool, likely safe, but not disclosed.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Single sentence, concise and front-loaded with purpose. No unnecessary words. Could optionally include more detail about output, but not required.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
The description covers the basic purpose and listed metrics. However, it doesn't mention output format (e.g., absolute differences, percentages) or that an output schema exists. Given the tool has an output schema, the missing details are less critical, but the description could be more complete.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The input schema provides descriptions for all parameters (e.g., 'Period A start (YYYY-MM-DD)'), so schema coverage is high. The tool description adds little beyond summarizing the metrics. Baseline 3 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states it compares two time periods across specific metrics (revenue, orders, etc.). This distinguishes it from sibling tools like compare_scenarios (which likely compares scenarios) or forecast_demand. However, it doesn't explicitly differentiate from siblings, so not a 5.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance on when to use this tool vs alternatives. For example, it doesn't mention that this is for comparing historical periods vs. compare_scenarios for hypotheticals. No exclusions or prerequisites provided.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Compare 2-4 promotional scenarios with what-if forecasting.
Each scenario specifies a promo period and discount depth. Returns a side-by-side markdown table with revenue projections, confidence bands, and a recommendation for the best-performing scenario.
| Name | Required | Description | Default |
|---|---|---|---|
| params | Yes |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries full burden. It does not disclose whether the tool is read-only, requires authentication, has rate limits, or has side effects (e.g., saving forecasts). It only describes the output, missing behavioral traits.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences, no redundancy. The first sentence states the core purpose, and the second adds input/output specifics. Every sentence earns its place, front-loaded with the main action.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the presence of an output schema, the description adequately covers inputs (scenarios with promo period and discount) and outputs (table with projections, confidence bands, recommendation). It could mention configurable parameters like horizon_days, but the schema provides those details. Almost complete.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 0% (no description of parameters in tool description), but the description summarizes that each scenario requires a promo period and discount depth. This adds some meaning beyond the schema, but it does not cover all parameters (e.g., horizon_days, context_days). Partially compensates for low coverage.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description states a specific verb ('Compare'), resource ('promotional scenarios'), and method ('what-if forecasting'). It clearly distinguishes from siblings like compare_periods by focusing on scenario comparison with discount and period inputs. The output format (markdown table with projections and recommendation) is also specified.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies usage when comparing 2-4 promotional scenarios, but it does not provide explicit guidance on when to use this tool versus siblings (e.g., compare_periods or analyze_promotion), nor does it mention when not to use it or any prerequisites.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Detect anomalous days where actual values fell outside expected forecast bands.
| Name | Required | Description | Default |
|---|---|---|---|
| params | Yes |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are present, so the description must fully communicate behavioral traits. It does not disclose whether the tool is read-only, requires specific permissions, or has side effects. The implication is that it only reads and returns results, but this is not explicit.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, well-structured sentence that is front-loaded and contains no extraneous information. Every word adds value.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
The description is too minimal given the tool's complexity (multiple parameters, output schema exists but not described). It fails to explain prerequisites, return values, or how 'actual values' and 'forecast bands' are sourced, leaving gaps for an AI agent.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The tool description does not mention parameters, but the input schema provides detailed descriptions for all four properties (lookback_days, sensitivity, metric, store). Thus, the schema already covers parameter semantics, earning a baseline score of 3.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's action ('detect anomalous days') and the condition ('where actual values fell outside expected forecast bands'). It is distinct from sibling tools that focus on forecasting, comparison, or promotion analysis.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance is provided on when to use this tool versus alternatives, prerequisites (e.g., existing forecast bands), or potential use cases. The description only states what it does without context.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Forecast demand by product, collection, or SKU using TimesFM 2.5.
Returns a ranked markdown table showing projected demand per group
with confidence bands. When group_value is 'all', forecasts the top N
groups by historical volume.
| Name | Required | Description | Default |
|---|---|---|---|
| params | Yes |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations provided, so description carries full burden. It mentions using TimesFM 2.5 model and returning markdown table with confidence bands, but does not disclose if it is read-only, requires permissions, data freshness, or side effects. For a forecasting tool, behavioral traits like 'safe to call repeatedly' or 'requires historical data' are missing.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences: first states overall purpose and output format, second clarifies a key parameter behavior. No wasted words, front-loaded with the core action.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Despite having output schema (not shown), description only covers group_by, group_value, and confidence bands. Missing explanation for lead_time_days, safety_factor, store, top_n, horizon_days, and metric enumeration. For a tool with 8 parameters and multiple sibling tools, this is incomplete.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema descriptions already cover all 8 parameters (group_by, group_value, metric, etc.) with defaults and ranges. Description adds value only for group_value='all' behavior. Baseline 3 due to high schema coverage.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Description clearly states verb and resource: 'Forecast demand by product, collection, or SKU'. It distinguishes from sibling forecast_revenue by mentioning multiple metrics (units, revenue, orders) and the use of TimesFM 2.5. The 'all' case for top N groups is also specified.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Provides some usage context: when group_value='all', forecasts top N groups by historical volume. However, no explicit guidance on when to use this tool vs siblings like forecast_revenue or analyze_promotion. No 'when not to use' or prerequisites stated.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Forecast total store revenue over a future horizon using TimesFM 2.5.
Returns a markdown summary and table with point forecast and 80% confidence band (q10-q90).
| Name | Required | Description | Default |
|---|---|---|---|
| params | Yes |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries the full burden but only mentions the return format (markdown summary/table) and model used. It does not disclose side effects, idempotency, or data handling, though the tool appears safe.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is extremely concise: one sentence specifying purpose and one sentence on output format, with no unnecessary words.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given an output schema is available, description covers return format but lacks error handling, edge cases, or prerequisites. It is adequate but not comprehensive.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Although the schema includes parameter descriptions, the tool description adds no additional meaning beyond the schema. Baseline 3 is appropriate given the schema coverage.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states that the tool forecasts total store revenue over a future horizon using TimesFM 2.5, which is specific and distinguishes it from sibling tools like forecast_demand.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance on when to use this tool versus alternatives like forecast_demand or detect_anomalies; no context on prerequisites or exclusions.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Identify seasonal patterns in store data by day of week, month, or quarter.
| Name | Required | Description | Default |
|---|---|---|---|
| params | Yes |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description must disclose behavioral traits. It only states the purpose, not how the tool behaves (e.g., read-only, aggregation method, data freshness, or side effects). The agent lacks critical context for safe invocation.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single, front-loaded sentence with no wasted words. Every part adds value, stating the action, resource, and key dimensions concisely.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the output schema exists and the input schema is well-described, the description covers the core purpose. However, it lacks any usage context or behavioral notes that would make it fully self-contained, so it is slightly above average but not complete.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The input schema provides descriptions for all four parameters (lookback_days, granularity, metric, store), so the description adds minimal extra meaning. It implies the granularity dimension but does not detail each parameter. Schema coverage is effectively high, so a baseline of 3 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Description clearly states the tool identifies seasonal patterns in store data by day of week, month, or quarter, specifying both the resource (store data) and the dimensions. It is distinct from sibling tools like forecast_demand or compare_periods, but does not explicitly differentiate them, preventing a 5.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance is provided on when to use this tool versus alternatives (e.g., forecast_demand for future predictions, compare_periods for period comparisons). There are no use case examples, prerequisites, or exclusions, leaving the agent to infer usage.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Each tool has a distinct purpose: analyze promotion impact, compare periods, compare scenarios, detect anomalies, forecast demand, forecast revenue, and get seasonality. No overlap in functionality.
All tools follow a consistent verb_noun pattern in snake_case (e.g., analyze_promotion, compare_periods, forecast_demand). No mixing of naming conventions.
Seven tools is appropriate for a forecasting and promotion analysis MCP server. It covers core needs without being too few or too many.
The tool set covers key forecasting, trend comparison, anomaly detection, and seasonality analysis. Minor gap: no tool for managing underlying data or listing available products/collections, but the surface is comprehensive for analysis.
Unclaimed servers have limited discoverability.
If you are the server author, to access and configure the admin panel.
Ask questions across Shopify, Klaviyo, GA4 and 20+ e-commerce sources in plain English.
Commerce intelligence for AI agents. Diagnose drop-offs, fix checkouts, optimize pricing.
Run storefronts, listings, orders, content, fulfillment, and analytics through AI.
All-in-one Shopify SEO. Rank on Google and in AI search. Find what's broken and fix it in chat.
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/mcostigliola321/shopify-forecast-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server
