rigging-agent

name

rigging-agent

description

Use whenever a used-machinery dealer needs to find a rigger / heavy-machinery mover for a specific location, or wants to add/update riggers in their personal rigger database. Pulls from the dealer's local rigger database (.dealer_riggers.json) first, then falls back to web search for riggers in the region that handle CNC, injection molding, presses, and other heavy industrial equipment. Triggers on "find a rigger", "who can rig in [city/state]", "rigger near [location]", "I need someone to load this", "add this rigger to my book", "rigger database", "save this rigger", or any time a dealer is sourcing the rigging side of a deal. NOT for freight/trucking — that's the trucking workflow.

Rigging Agent

Sources riggers / heavy-machinery movers for a dealer's loads. Two modes: pull from the dealer's personal book (the fast, trusted answer) or search the web for new candidates in regions the dealer hasn't worked before.

When this fires

• "Find me a rigger in Cleveland for the KraussMaffei"
• "Who do I have in Texas for a 50,000 lb press?"
• "I need a rigger near Columbus, OH — injection molding, dock-high"
• "Add Bayshore Rigging to my list — [phone], Cleveland, IMM specialist"
• "Pull up my Ohio riggers"
• "Search for riggers near 46815 that handle CNC"

First-run setup

If /mnt/user-data/uploads/.dealer_riggers.json doesn't exist, create it as an empty list:

{
  "riggers": [],
  "preferred_radius_miles": 250,
  "last_web_search_cache": {}
}

Rigger record schema

Each rigger in the database:

{
  "id": "bayshore-rigging-cleveland",
  "company": "Bayshore Rigging",
  "contact_name": "Mike Anderson",
  "phone": "[phone]",
  "email": "[e-mail]",
  "website": "https://bayshore-rigging.example",
  "city": "Cleveland",
  "state": "OH",
  "postal": "44102",
  "service_radius_miles": 300,
  "equipment_types": ["injection_molding", "cnc", "presses", "heavy_general"],
  "capabilities": ["crane_to_50t", "dock_high", "millwright", "trucking_partner"],
  "last_used_date": "2026-02-14",
  "last_used_load": "Krauss-Maffei CX 160 load-out",
  "rating": 5,
  "notes": "Sharp, on-time, sent photos during load. Will quote teardown.",
  "added_by": "dealer",
  "source": "personal"
}

source is "personal" (dealer added) or "web" (skill found and dealer approved). equipment_types enum: injection_molding, cnc, presses, extrusion, blow_molding, robots, heavy_general. capabilities is free-form but pick from common: crane_to_50t, crane_50_100t, crane_100t_plus, dock_high, forklift_to_25k, forklift_25k_plus, millwright, disassembly, trucking_partner, crating, international.

Per-request flow

1. Parse the ask — extract location (city/state/zip), equipment type, and any constraints (weight, dock height, crane size, timeline).
2. Query local DB first — call scripts/find_riggers.py --location <loc> --equipment <type>. Returns matches ranked by:
   • In-state and within service_radius_miles of the load location
   • Equipment-type match
   • Recency (riggers used in last 12 months ranked higher)
   • Rating
3. Present matches — show top 3 with contact info, last-used date, and the load they ran. Include the dealer's own notes verbatim — those notes are the real value.
4. If no local matches (or dealer asks to broaden) — scripts/web_search_riggers.py writes a search-results stub; the skill then uses the WebSearch tool to find candidates with queries like:
   • "machinery rigging" near "<city>, <state>"
   • "injection molding mover" <state>
   • heavy machinery riggers <state> CNC press For each web result the skill collects: company, phone, email, website, location, equipment types mentioned. Present the candidates and ASK the dealer which to call. Do NOT auto-add web results to the database — only add after the dealer says "use this one" or "add them" or after they actually run a load with them.
5. Saving / updating — scripts/add_rigger.py adds or updates a record. Always confirm with the dealer before writing: "Adding Bayshore Rigging (Cleveland, OH, IMM specialist) to your book — confirm?"

After a load runs

When the dealer mentions a load completed with a rigger ("Bayshore loaded the KraussMaffei yesterday"), update that rigger's last_used_date, last_used_load, and prompt for a rating + one-line note. The notes compound into real institutional knowledge.

Anti-patterns

• ❌ Auto-saving web-search results before the dealer confirms — the database is curated, not a dump.
• ❌ Recommending a rigger without surfacing their last-used date and the load context — those are the trust signals.
• ❌ Forgetting to filter by equipment_types — a generic heavy-hauler is not the same as an IMM specialist.
• ❌ Searching the web before checking the local DB — the dealer's own book is always first.
• ❌ Conflating riggers with truckers — trucking_partner: true means they can hand off to a trucker, not that they ARE one.

File map

• scripts/find_riggers.py — local DB query (location + equipment filter, ranked output)
• scripts/add_rigger.py — add/update a rigger record (interactive prompts via skill, batch-safe)
• scripts/web_search_riggers.py — emits a search stub; pairs with the WebSearch tool to find candidates
• references/rigger_schema.md — full field reference and enum values

Companion skills

• sales-proposal — when rigging_mode="included", the rigger named here goes on the proposal
• past-sales-agent — every closed load logs which rigger ran it; that data feeds back into rigger ratings