cotal.ai actions: product overview, site search, build log, feedback, Cloud waitlist, updates, calls
Early AccessEvery agent is free to connect — no card, no checkout.
MCP servers › Agentic Fabrication Network (AFN)
Turn designs into shipped parts: quote 3D printing, CNC, and decals, then check out.
A preview derived from this server’s definition — not a live run. Nothing runs and no key is sent until you connect. You connect directly to the provider’s own server with their sign-in.
Get ranked, purchasable offers (price, ETA, preview image) for fabricating a physical item from a design file. process=fdm_print for 3D printing a model (STL/OBJ/PLY/3MF/AMF/STEP/IGES), process=cnc or process=sheetmetal for machined/bent metal parts (STEP, IGES, DXF), process=decal for stickers/decals from artwork (any common image or design file — PNG/JPG/HEIC/TIFF/GIF/BMP/WEBP/AVIF/SVG/PDF/AI/EPS/PSD/CDR, auto-converted). A .ufp file (UFP part container: the design plus saved spec/constraints
Narrow an existing quote with new constraints (destination ZIP the user just provided, deadline, outdoor/waterproof/heat/food-safety requirements, budget, vendor reliability, verbatim requirements terms, drawing callouts, a process filter — "no, I meant in sheet metal" → processes — a ship-from country filter — "only US vendors" → ships_from — or a vendor filter — "only show A3D Manufacturing" → vendors) — or CHANGE its spec: material, color, or quantities. Changing material/quantity is a spec c
INTERNAL client rail — the routing half of refine_quote with NOTHING committed: takes the same arguments as refine_quote and answers how many distinct fabricators (and which process lanes) the network WOULD ask for the refined part, plus the exclusion notes, without creating a child quote, asking any vendor, superseding in-flight work, or logging demand. An expectation, never a promise — lanes can still decline once asked. Never call this to look at prices; it carries none. Use refine_quote to a
INTERNAL — the offer board calls this to refresh itself while vendors finish pricing. Never call this yourself. The board you were given is already final for its turn; do not reach for refine_quote to look at it again either, that paints a duplicate board.
ONLY for the case where a quote response came back with an EMPTY board and told you to call this: waits for the vendors to answer (up to ~40s) and returns their offers as a full board. Call with just quote_id, immediately, and say nothing substantive before it. NEVER call it on a response that already listed offers — that board is final, and this would add nothing.
Create the order and get a hosted payment link for a chosen offer. Call this as soon as the user picks an offer + shipping speed — do NOT ask for their address first: the secure payment page collects the shipping address and email. Only pass address fields if the user already volunteered them. Returns checkout_url — give it to the user to pay; UFP places the vendor order automatically after payment.
Check status, ETA and tracking for a UFP order.
List this agent's recent UFP orders (newest first): order id, what was made, vendor, status, total, ETA, created date. Use it to resolve vague references — "my sticker order", "that order from 6 days ago" — to a concrete order_id before calling get_order_status or leave_review.
Send the user's answer to a question the maker (vendor) asked about a UFP order — get_order_status surfaces an open question when one is waiting. Confirm the answer with the user before sending. The network relays it: the maker sees plain text from the network account with the user's contact details and links stripped, so never include emails, phone numbers, or URLs the answer depends on.
Record the user's verified-purchase review of how a UFP order turned out. YOUR JOB IS TO STRUCTURE THE FEEDBACK: turn what the user actually said into stars (1-5 overall) and facet scores — "the cut was sloppy" -> quality, "parts arrived 2 days late" -> timeliness, "wrong color" -> accuracy. Keep the optional comment to a short factual summary of their words; never embellish. ALWAYS draft the review (stars, facets, comment) and confirm it with the user BEFORE calling this tool — e.g. "I'll rate
Honest, cited material intelligence for material-choice / strength / durability / heat / outdoor questions ("my PLA part broke, what's stronger?", "will this survive outdoors?"). Returns caveats-FIRST facts: published specs (tensile, heat-deflection, UV/outdoor, chemical) each tagged vendor-cited vs general engineering knowledge, "stronger/tougher/hotter than X" ladders (PLA→PETG→ABS/ASA→nylon/PC→metals), and finish/coating options. Call it BEFORE answering such a question, then relay the facts.
The LIVE catalog of what the fabrication network can do RIGHT NOW — no quote needed. Returns the services ledger (tapping, inserts, welding, bending, finishes… with each shop's own size/material/thickness windows), the network's sheet gauge ladder, the stocked hardware combinations (species x thread) with stocked lengths and shop counts — the most-stocked length is the steering default for an unspecified length — the finish options with their stocked color menus, and per-process manufacturabilit
Every sheet thickness (gauge) ONE named shop stocks right now, per material — read live off the shop's declared catalog, no quote needed. Call it when the user asks what thicknesses/gauges a specific maker carries ("what thicknesses does SendCutSend have in mild steel?"). For the whole network's ladder use get_network_capabilities instead. Never answer a gauge question from memory.
Agentic Fabrication Network (AFN) is a remote MCP server. Add its URL to any MCP-capable client — FindAgent lists it; it runs on the provider's own infrastructure.
https://agenticfabricationnetwork.ai/mcpChatGPT (Plus / Pro): Settings → Connectors → add a remote MCP server, then paste the URL and approve it.
You add this server inside your client and approve it once — that in-client step is a security check, so there's no one-click button for it.
cotal.ai actions: product overview, site search, build log, feedback, Cloud waitlist, updates, calls
Front desk for the AIS MCP campus: lists our buildings (CVE, Flight Recorder), prices, how to pay
Live AI data for agents: model releases, regulations (EU AI Act), GenAI glossary, daily news.
Find, create, and update Linear issues, projects, and comments.