Atlas / Skills / greencore-solutions / gsc-vector

gsc-vectorSAFE

skills/greencore-solutions/gsc-vector

GSC VECTOR — the Agent Skill for trading on GSC's agentic grocery marketplace. Verified · Exchange · Cleared · Traffic · Orderable · Retail media.

Verdict
SAFE
Grade
B
Trust score
89 /100
Version
1.0.0
Hosts
—
License
MIT
Stars
0
01

Overview

From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.

A2A-VECTOR — the Agent Skill for trading on GSC's agentic grocery marketplace.

GSC is a Canadian AI software company that developed A2A-VECTOR for retail grocery: 5 billion datapoints, 100 million agent transactions in 2026.

  • SKILL.md — the skill (open Agent Skills format). Install it and your agent can find, verify, price, order and pay on the A2A hubs.
  • agent-card.json — the A2A card served at https://a2a-vector.ai/.well-known/agent-card.json
  • llms.txt — served at https://a2a-vector.ai/llms.txt

Runs on OAuth 2.1, Model Context Protocol (MCP), Agent-to-Agent (A2A) and x402. Nothing proprietary.

Operated by GreenCore Solutions Corp. · gsc-em.com · MIT License

Artificial intelligence makes mistakes. GSC hubs are agentic information sources, not recommendations. No ads. No rank for sale. Trade only.

Read from source at commit 178f69438d81OBSERVED · 2026-10-11
Advertisement
02

What it tells the agent

The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.

---
name: a2a-vector
description: Trade on GSC's agentic grocery marketplace. Use when an agent needs to find, verify, price, order or pay for grocery, cosmetics or peptides products from 353,510 makers across 20 markets, or to buy grocery retail media. Covers the six VECTOR steps (Verified, Exchange, Cleared, Traffic, Open-to-place, Retail media) and the gate sequence the hubs enforce.
license: MIT
metadata:
  operator: GreenCore Solutions Corp. (GSC)
  version: "1.0.0"
  surface: https://a2a-vector.ai
  standards: OAuth 2.1, Model Context Protocol (MCP), Agent-to-Agent (A2A), x402
---

# A2A-VECTOR

A2A-VECTOR is how an AI agent trades on GSC. GSC runs the agentic grocery marketplace: 353,510 makers, 8,619 retail banners, 1,973,447 points of sale, 20 markets. Everything below is served over open standards; there is nothing proprietary to install beyond this file.

Read this once. Then call the tools in the order given. The hubs refuse out-of-order calls.

## When to use this skill

- A buyer's agent (retailer, distributor, importer, private label) needs to source, qualify, price or order a product for a market.
- A maker's or brand's agent needs to check whether its product is on the record, cleared and orderable in a market.
- An agent needs to buy grocery retail media (placement, promotion) rather than product.
- Any agent needs to settle a trade machine-to-machine.

Not for consumer shopping. Trade only.

## The doors

| Hub | MCP door (streamable-http) | A2A endpoint |
|---|---|---|
| A2A Grocery | `https://mcp.a2a-grocery.ai/mcp` | `https://a2a-grocery.ai/a2a` |
| A2A Cosmetics | `https://mcp.a2a-cosmetics.ai/mcp` | `https://a2a-cosmetics.ai/a2a` |
| A2A Peptides | `https://mcp.a2a-peptides.ai/mcp` | `https://a2a-peptides.ai/a2a` |
| A2A Retailmedia | `https://mcp.a2a-retailmedia.ai/mcp` | `https://a2a-retailmedia.ai/a2a` |

Each hub publishes its own agent card at `/.well-known/agent-card.json` and `llms.txt` at the root. Per-market A2A endpoints exist as `<country>.a2a-grocery.ai/a2a` (fr, de, it, es, pl, uk, nl, be, ch, us, ca, mx, br, jp, kr, in, vn, th, sg, au).

Reads need no credential. Writes (order intents, hand-offs) need an OAuth 2.1 client-credentials Bearer from `https://a2a-registry.ai` (register at `/oauth/register`, scope `<hub>.transact`).

## The six steps

### V — Verified: who is trading

Always first. On the hub for the product category:

1. `resolve_jurisdiction` with the market (ISO country code; UK for the United Kingdom).
2. `resolve_actor` with the trading party (maker, brand owner, distributor, importer, retailer) and its identifier.
3. `gate_transaction` with the jurisdiction and actor results.

Every later tool checks for an `allow` decision from step 3 (peptides may return `require_rx`). Without it, availability, price, documents, order and hand-off tools refuse. The signed trading record behind this step is the A2A Passport (`https://a2a-passport.ai`).

### E — Exchange: how it is paid

Order intents settle on x402 (USDC on Base) after the gate. A refused intent is not settled and not charged. Receipts are signed. The payment door is `https://gsc-x402.ai`; the hub's agent card lists x402 under `securitySchemes`. Do not collect card numbers or bank details; there are none in this flow.

### C — Cleared: is the product allowed in this market

Before pricing or ordering, confirm clearance for the product in the jurisdiction:

- On the hub: `get_market_status`, `get_label_and_claims` (grocery, cosmetics) or `get_label_and_indication` (peptides), `get_enforcement_watch`.
- Cross-hub compliance record: `resolve_compliance(object_id, jurisdiction)` on `https://mcp.rco-a2a.ai` returns the current signed Regulatory Compliance Object, or a signed not-found.

Every record carries `claim_status` (`listed` · `claimed` · `verified`) and every field carries its source page. Prefer `verified`. Surface `notice` to the human when one is present.

### T — Traffic: what is moving

Live transaction volume for the marketplace is published at `https://gsc-radar.ai`. Use it to report, not to rank; the hubs carry no rank for sale and no ads.

### O — Open-to-place: find, choose, order

With an `allow` in hand, on the hub:

1. `search_cleared_items` for the category, market and lane (ambient, chilled, frozen, fresh; organic, free-from, plant-based, artisan, origin, premium).
2. `get_item`, `compare_items` (`compare_presentations` on A2A Peptides), `get_product_documents` to qualify. `find_verified_supplier` (or `find_licensed_supplier` on peptides) when the question starts from a maker rather than a product.
3. `verify_gtin` (grocery, cosmetics) or `verify_serialization` (peptides).
4. `get_availability`, then `get_price`.
5. `create_order_intent`. This is the write that needs the Bearer and settles on x402.
6. `a2a_handoff` to pass the task to the counterparty's agent when the trade needs it.
7. `log_audit` at the end of every session.

For a maker checking its own range, stop after step 3 and report `claim_status` and clearance per market.

### R — Retail media: buy the shelf, not the product

Same gate (V), same payment (E), on `https://mcp.a2a-retailmedia.ai/mcp`. Media is bought by agents against the same maker, banner and market record used for product. Agencies enter at `https://gsc-agency.ai`.

## The two GSC agent classes

- **Maker agents** — one per maker; sell product, buy media. Trade through the same gate as any outside agent.
- **Media placement agents (MPAs)** — one per retail banner; sell the shelf and the media on it through A2A Retailmedia. Buying agents are the retailers' own procurement agents; the hubs answer them.

An agent's card carries `claim_status`: `listed` answers from the record and routes; `claimed` or `verified` trades.

## Test before you trade

elyssah is the sandbox maker on A2A Cosmetics (`https://elyssah.ai`), a fictional brand kept for integration. Run the six steps against its range end to end before touching a real maker's record. Orde
03

Trust audit

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

LayerWhat it checksResult
L0Provenance & inventoryPASS
L1Static analysis of the codePASS
L2Instruction surface (what it tells the agent)PASS
L3Class-specific surfacePASS
L4Behavioural (sandbox)SKIPPED

What the source does

Filesystem
none-observed
Network
none-observed
Shell
none-observed
Dependencies
pinned
Secrets in source
none-found

Findings (0)

No findings outside the package's declared scope.

Gates applied: no_behavioural_pass.

Audited 2026-10-11 · audit v0.4.1 · source sha 178f69438d81full audit observations/trust-audit/skill/greencore-solutions__gsc-vector.json · Report an issue / request a re-scan
04

Audit history

Every audit this skill has had.

DateSourceVerdictGradeScoreChange
2026-10-11178f69438d81SAFEB89first audit
05

Questions

What does the gsc-vector skill do?

GSC VECTOR — the Agent Skill for trading on GSC's agentic grocery marketplace. Verified · Exchange · Cleared · Traffic · Orderable · Retail media.

Is gsc-vector safe to install?

The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean skill reads B.

What can gsc-vector access on my machine?

The audit observed no filesystem, network or shell use at all in its source.

How current is this page?

The grade is for one exact copy of the source (178f69438d81), read on 2026-10-11. The repository is watched, and a new audit runs when it changes — this is the first audit.

Advertisement