InfiPlex Knowledge Base

Two InfiPlex reports run daily on their own; every other report can run on a schedule through the MCP server with an AI agent or the REST API with your own script, and land in your team's inbox, Slack, or shared folder every morning. This guide is the decision page: what runs automatically today, when to choose the agent route over a script, which MCP tool or API endpoint feeds each common report (inventory by warehouse, low stock with velocity, SKU plus secondary ID, orders by channel, shipped with tracking, failed channel calls), how delivery works, and the ready-built agent guides to copy. Read scopes only, nothing metered beyond API access.

A copy-ready agent guide: every evening an AI agent on the InfiPlex MCP server calls Warehouse Search once, pages through Inventory Search, pivots the rows to one line per SKU with a column per warehouse and a total, writes the CSV and a short summary, and the run script emails both to your distribution list. Read-only key with Inventory and Warehouse scopes, so the agent cannot change a quantity. Includes the instruction file, the MCP config, the run script with the mail line, the cron line, a sample sheet and summary, and variants for adding a secondary ID or supplier code column, one warehouse only, changes only, and per-client runs on a 3PL account. About 25 tool calls for a 1,200-SKU, two-warehouse catalog.

A receiving sheet with three columns, SKU, incoming quantity, and warehouse name, goes into InfiPlex as additions to stock after one preview and the word go. An AI assistant on the InfiPlex MCP server maps warehouse names to IDs, builds one quantity_to_add row per line, and calls inventory_set_bulk, which returns a preview and a server-issued token before anything is written; unresolved rows are listed, not guessed. This guide covers the key to create, the sheet format, the chat session with a sample preview, the same six rows as a plain Inventory Set Bulk API call for anyone who prefers a script, and the unattended version that the autonomous fulfillment agent runs from a folder every night. Every executed row lands in the SKU's inventory history under the receiving key.

SKU Data is the per-SKU key and value store in InfiPlex. Reserved names such as thehomedepot_sku and costco_sku are read by connections; any other name is a custom entry InfiPlex keeps for you: a legacy item ID, a supplier code, a case pack, a warranty length. This guide covers the reserved-vs-custom rule and the misspelling trap, loading values by file through SKU Utilities or by API through SKU Data Set and Bulk Update, reading them back flat with SKU Data Get or pivoted with SKU Data Search, the two-line change that puts a secondary ID on the daily inventory by warehouse report, and the lookup that turns an old item number into the SKU and its on-hand count, in a script or in chat on the MCP server.

A marketplace InfiPlex does not connect to yet still runs on InfiPlex inventory and ships from the InfiPlex order list; three hand-offs happen by file instead of by API until the connection is built. Inventory out: the Price & Inventory Download mapped into the marketplace's upload, or a nightly read-only agent on the MCP server that writes the marketplace's stock file with a hold-back. Orders in: Manual Order Import, SFTP order files, or the Order Create API, with every imported order drawing down stock for all other channels. Tracking back: the Shipping Item Report scoped to that order source, or an agent writing the tracking file from each order's packages. Then the path to the real integration: Growth Packages include building API-based marketplace connections at no additional cost, and InfiPlex Bridge covers EDI-only retailers.

A scheduled AI agent that builds the day's ship queue from InfiPlex before the first shift: every open order not fully shipped, grouped by channel and sorted oldest first, with hours since it arrived, the warehouse each line routed to, and a LATE or DUE flag against that channel's handling window (Amazon Direct Fulfillment, Walmart DSV Expected Ship Date, Temu handling time, Seller Fulfilled Prime). For flagged orders it reads the warehouse routing explanation to show whether stock was short at routing. On a key with Write Orders it sets flagged orders to needs review with a note. Built on the InfiPlex MCP server with a Read Orders and Read Warehouse key, Claude Code in headless mode, and cron. Includes the window table with the rule each comes from, key settings, full instruction file, run script, sample queue, and checklist.

A scheduled AI agent for the InfiPlex shop log. Every morning it calls log_types_get, keeps every Failed type, searches each for the last 24 hours, groups failures by channel and cause, attaches the current state of every affected order, and checks the channel's success type after the last failure to mark each group ongoing or recovered. On a key with Write Orders it also adds an internal note to affected orders. Built on the InfiPlex MCP server with a Read Log and Read Orders key, Claude Code in headless mode, and one cron line. Includes how log types work, both key versions, the full instruction file, run script, schedule, a sample report showing Walmart DSV and Temu tracking sends that never reached the channel, and the checklist. It reports; repairs such as re-recording tracking stay previewed writes on a warehouse key.

A scheduled AI agent that reads InfiPlex every night on a read-only key and leaves a reorder sheet for the buyer: every SKU at or below its reorder point, units sold over the last seven days from the change history, stock in each warehouse, days of cover, and a suggested order quantity built from lead time, safety stock, and your order cycle. Uses the InfiPlex MCP server and four read tools (inventory_restock_list, inventory_change_history_get, inventory_get, warehouse_search), so nothing can change and there is no confirmation flow. Includes the key settings, the full instruction file, the Claude Code run script and cron line, a sample CSV and summary, and how to tune the suggested quantity. The same reorder point drives the Restock Alerts email; this adds how low, how fast, and where. Included with API access on every InfiPlex plan.

This guide sets up an AI agent that runs InfiPlex fulfillment on a schedule with no person in the loop. Every morning it applies the warehouse's receiving file in one batch, records the carrier's tracking export against the right orders so tracking reaches each marketplace, lists anything that fell to its reorder point, scans every channel's failed log type for the last 24 hours, and writes a plain-text report your team reads at 8 AM. It runs on the InfiPlex MCP server with a dedicated fulfillment API key whose scopes bound everything the agent can do, a standing instruction file that authorizes the agent to complete the server's preview-and-confirm flow itself, and Claude Code in headless mode launched from cron with tool, turn, and spend caps. The guide includes the exact key configuration, the full instruction file, the MCP config and run script, a read-only dry-run method, a sample morning report, the kill switch, and how to widen the agent's job one scope at a time.

The InfiPlex MCP server lets you connect Claude, ChatGPT through Codex, Cursor, or any MCP-compatible AI agent directly to your own InfiPlex instance, so your team can search orders, explain warehouse routing, check inventory, read the shop log, record tracking, and set prices in plain English. This guide covers the whole set-up in the order you will do it: opening or creating an API key, choosing which read and write scopes the agent should have, enabling MCP and understanding the two MCP flags that gate setting changes and live marketplace traffic, and then connecting each client with the two values shown on the key page. Separate walkthroughs cover Claude on claude.ai and Claude Desktop, Claude Code, ChatGPT via Codex, and generic MCP clients such as Cursor and Windsurf. It closes with five first questions that confirm the connection and the preview gate, a troubleshooting list keyed to the server's exact messages, and a set-up checklist.