plm-upload-mcp

An MCP server for the LionsBot PLM. It reads the PLM — SKU folders, full component metadata, extracted drawing/BOM metadata, and product & staging BOMs — and bulk-uploads local files (SolidWorks STEP/PDF exports) into it, matched to a component by SKU. It runs on your own machine, so it reads your exported files straight off disk, and it signs in as you with your PLM/hub login — every read is row-level-security scoped to you, and every uploaded file is tagged with who uploaded it.

Pair it with your SolidWorks MCP: that MCP knows each part's SKU and exports the files; this MCP takes {path, sku} pairs and pushes them into the PLM.

Setup

Add one entry to your Claude MCP config (Claude Desktop: claude_desktop_config.json; Claude Code: .mcp.json or claude mcp add). Use your own PLM email + password:

{
  "mcpServers": {
    "plm-upload": {
      "command": "npx",
      "args": ["-y", "plm-upload-mcp"],
      "env": {
        "PLM_EMAIL": "[email protected]",
        "PLM_PASSWORD": "your-plm-password"
      }
    }
  }
}

That's all. npx -y fetches and runs the latest version; no clone, no build.

Your login is the same one you use for the Hardware Tools hub / PLM. Uploads are attributed to you and require a @lionsbot.com account.

Tools

upload_files

Uploads one or more files, each paired with its target SKU.

• files: array of { path, sku, skuName?, category? }
  • path — absolute local path to the file
  • sku — component SKU, e.g. MNT-1234-A0
  • skuName — optional; only used if the SKU's Drive folder has to be created
  • category — optional PLM tag: Default | Archive | Compliance | Ops
• bucket: main (default, confirmed components) or staging (unconfirmed)

Each file lands in PLM storage and is mirrored into that SKU's Google Drive folder (created automatically if it doesn't exist). Re-uploading a file with the same name replaces it in place — no duplicate pile-up on re-runs.

Example ask: "Upload C:\exports\bracket.step to SKU MNT-1234-A0 and C:\exports\bracket.pdf to the same SKU."

list_component_files

Lists the files currently in a component's Drive folder.

• sku: component SKU
• bucket: main (default) or staging

Read-only tools

All read tools query the PLM as you (RLS-scoped) — you only ever see what you're allowed to see. bucket / classification values are main (confirmed), staging (unconfirmed), or all (default).

list_sku_folders

Every component SKU with its Google Drive folder — the SKU→folder map (SKU, name, category, type, revision, status, staging flag, folder id/URL). Filters: bucket, category, with_folder_only, limit, offset.

get_component_metadata

Full component metadata (every column, incl. cost/supply, drawing-extracted fields, CAD provenance, compliance, and the extract_meta block). Pass a single sku for one part, or omit it to page through all. Filters: search, bucket, category, status, limit, offset.

list_bom_extract_metadata

The metadata machine-extracted from each part's drawing/BOM PDF (extract_meta): the Lionsbot-custom flag, SKU/drawing match check, per-field extracted values, and source file + who/when. Only extracted parts are returned. Filters: sku, bucket, limit, offset.

list_boms

BOMs across all products, each product carrying its BOM(s) + line items and a derived staging flag (a product is staging while its working BOM still contains any staging component). Working BOMs only by default. Filters: product_sku, classification, include_snapshots, include_rows, max_rows_per_bom.

Config overrides (rarely needed)

Baked to production defaults; override via env only if the PLM backend moves:

| env | default | |---|---| | PLM_GATEWAY | https://gateway-production-c13d.up.railway.app | | PLM_ANON_KEY | (prod anon key) | | PLM_EDGE_URL | https://plm-edge-production.up.railway.app |

Requirements

Node.js 18+ (for built-in fetch).