MCP CPG Agent Protocols

CPG Agent Protocols

mcp.cpgagentprotocols.ai — the standards door of the GSC AI Agent Stack. A stateless standards beacon for ACM-SPARKS, SM-ECO-10060 and ACM-68000.

The beacon serves the protocols themselves, compiled as canon: ACM-SPARKS classifies the SKU, SM-ECO-10060 resolves the jurisdiction, ACM-68000 emits the deterministic signal to the buyer's AI Agent. No database, no secrets, no state; every answer is the same for every agent that asks.

Canonical endpoint: https://mcp.cpgagentprotocols.ai/mcp (streamable-http, stateless).

This repository holds the beacon's source (server.js, canon.js, DEPLOY.md) and this connect kit: the tool list as the live server states it, the signal register, and one runnable example per language against the live endpoint. Nothing here carries a key, a secret or a customer.

Tools

As returned by tools/list on the live beacon on 2026-09-09.

Tool	What it answers
list_protocols	The protocol family and the spine: ACM-SPARKS classifies, SM-ECO-10060 resolves, ACM-68000 signals.
get_protocol	Full canon record for ACM-SPARKS, SM-ECO-10060, or ACM-68000.
list_signals	The signal register, enumerated individually. One meaning, no drift.
resolve_signal	The state and meaning of a single ACM-68000 signal code. A CPG- profile code resolves to its ACM- twin; the response carries the resolved ACM- code.
list_sparks_dimensions	SPARKS resolves six dimensions — SKU, Pack, Amount, Region, Kernel, and Standard — a CPG industry first.
resolve_sparks_dimension	Canon record for a single SPARKS dimension by letter or name.
list_members	The sovereign jurisdictional member nodes: code, namespace, jurisdiction, region, and GS1 territory GTIN.
resolve_member	Canon record for a single sovereign member by code, e.g. FR, AU, MX.
get_header_canon	The GSC wire-header canon: the x-gsc-* headers every GSC surface emits, in canon order.

The signals

The register. Each ACM- signal has a CPG- profile twin naming the same definition.

ACM- (wire form)	State	CPG- profile	Meaning
acm-000	NOT_APPLICABLE	cpg-000	Control or framework does not apply to this SKU.
acm-200	ALLOW	cpg-200	Eligible for autonomous procurement. Buyer agent may complete the order with human-in-the-loop confirmation in the retailer's chosen ERP.
acm-300	CONDITIONAL	cpg-300	Eligible subject to a documented condition (volume threshold, jurisdictional carve-out, time-bounded approval).
acm-403	RESTRICT	cpg-403	Not eligible in the requested context. Buyer agent must not complete the order.
acm-404	NOT_FOUND	cpg-404	GTIN is not registered. No signal available.
acm-451	ESCALATE	cpg-451	Cannot be resolved deterministically. Human review required. Served by mcp.cpghumanintheloop.ai.
acm-500	SYSTEM_ERROR	cpg-500	Server or upstream error. Retry per backoff policy.

Any of the CPG- codes in the table resolves to its ACM- definition on the same row; the wire form is ACM-. The beacon's resolve_signal accepts either form and answers with the ACM- record, the resolved ACM- code on the response (code) beside the code as given (input_code); anything outside the seven is NOT_FOUND.

Links: https://acm-68000.ai · https://cpg-68000.ai · https://standard-10060.org

Examples

Walk	Python	TypeScript
call the beacon: tools/list, then list_protocols, list_signals, list_sparks_dimensions; print the protocol set it returns; stop	examples/beacon.py	examples/beacon.ts

Python 3 standard library only. TypeScript runs on Node 24 as-is (node examples/beacon.ts) with no dependencies. Live endpoint only.

Sources on the wire

https://cpgagentprotocols.ai · https://mcp.cpgagentprotocols.ai/protocol.json · https://mcp.cpgagentprotocols.ai/signals.json · https://mcp.cpgagentprotocols.ai/sparks.json · https://mcp.cpgagentprotocols.ai/nodes.json · https://acm-sparks.ai · https://sm-eco-10060.org · https://acm-68000.org

Registry listing: io.github.greencore-solutions/cpg-agent-protocols.

---

GreenCore Solutions Corp. — 4611 Viking Way, Suite #260, Richmond BC V6V 2K9 Canada GSC Agentic Pty Ltd — Level 1, 63-73 Ann Street, Surry Hills NSW 2010, Australia

Operated by GreenCore Solutions Corp. · https://gsc-em.com · @GSC_Rail_ai · @ACM68000