global-support-standard 0.2.2

GSS Python reference implementation: provider API + CLI consumer

Global Support Standard (GSS)

Open protocol for machine-readable e-commerce support.

GSS lets any app, AI agent, or device resolve customer support requests directly with a shop using a consistent command model:

gss <shop> <domain> <action> [--flags]

This repository includes a Python reference implementation of the protocol and supporting project docs.

Why GSS

Most support requests are policy execution, not free-form conversation:

• "Where is my package?"
• "Can I return this item?"
• "Why is my refund delayed?"
• "Can you show my account audit trail?"

Today these flows are reimplemented per shop and routed through human agents. GSS standardizes these operations so support can be:

• Faster (seconds, not days)
• Consistent across shops
• Safer for automation (action levels, confirmation tokens, auditability)
• Consumer-agnostic (AI, mobile app, browser extension, device)

Repository At A Glance

Area	Purpose
src/gss_provider	FastAPI provider server (shop-facing implementation)
src/gss_cli	Typer CLI consumer (gss command)
src/gss_core	Shared models, envelope helpers, and error contracts
protocols/	Protocol format docs and runnable YAML examples
providers/mock_shop	Example consumer-policy configuration
sdk/typescript	TypeScript SDK scaffold (non-blocking roadmap)
spec/	Protocol/spec narrative and architecture references
docs/	Getting-started docs for shops and consumers
tests/	API and CLI integration tests

Reference Implementation Scope

This codebase currently ships an end-to-end production baseline:

• Provider API with standardized response envelope
• CLI with gss <shop> <domain> <action> routing
• Local demo adapter/data layer
• Protocol engine (YAML rules + context enrichment)
• Security baseline:
  • Customer auth token flow
  • Required GSS-Consumer-* headers
  • Two-step request execution (returns initiate -> returns confirm)
  • Append-only audit log records
• Stateless core boundary:
  • Framework defines contracts and orchestration
  • Shops own persistence/secret storage in adapter implementations

Architecture

flowchart LR
  consumerClient[Client app/AI/device] -->|gss commands| cli[gss CLI]
  cli -->|HTTP + GSS headers| provider[gss_provider FastAPI]
  provider --> describe[DescribeService]
  provider --> domains[Domain Services]
  provider --> protocols[Protocol Engine]
  provider --> auth[Auth Context]
  provider --> audit[Audit Logger]
  domains --> adapter[Reference Shop Adapter]
  protocols --> yaml[Protocol YAML files]

Quickstart

1) Install

python -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"

2) Start provider

gss-provider

Default endpoint: http://127.0.0.1:8000/v1

3) Run consumer flow

gss mockshop.local describe
gss mockshop.local auth login --method api_key --customer-id CUST-001
gss mockshop.local orders list
gss mockshop.local orders get --id ORD-1001
gss mockshop.local shipping track --order-id ORD-1001
gss mockshop.local protocols get --trigger delivery-not-received --context '{"order_id":"ORD-1002","days_since_expected":1}'
gss mockshop.local returns initiate --order-id ORD-1001 --item-id ITEM-1 --reason defective
gss mockshop.local returns confirm --token <confirmation_token>
gss mockshop.local account audit-log

Optional token behavior:

• default: CLI stores token locally in ~/.gss/tokens.json
• disable local storage: GSS_STORE_TOKENS=0
• provide token explicitly: GSS_CUSTOMER_TOKEN=<token>

Shopify webshop project (inside this repo)

gss-shopify-provider

See webshop/shopify-test-store/README.md for setup against your test store. The Shopify demo includes an agent-first auth flow (auth verify-customer -> auth issue-token) in addition to legacy dev login.

HTTP Endpoints

Discovery

• GET /v1/describe
• GET /v1/{domain}/describe

Auth

• POST /v1/auth/login

Domains

• GET /v1/orders
• GET /v1/orders/{order_id}
• GET /v1/shipping/track/{order_id}
• POST /v1/returns/check-eligibility
• POST /v1/returns/initiate
• POST /v1/returns/confirm
• POST /v1/protocols/get
• GET /v1/account/audit-log

Security Model Highlights

• Every protected request requires:
  • Authorization: Bearer <token>
  • GSS-Consumer-Id
  • GSS-Consumer-Type
  • GSS-Version
• request actions are two-step by design (issue confirmation token, then confirm)
• Authorization is customer-scoped and domain actions enforce ownership checks
• Request/action events are recorded to the audit log
• Trust signaling:
  • describe includes compliance metadata (level, certified, test_suite_version)
  • CLI warns when a shop is uncertified or missing compliance metadata

Production Minimum Config

For production deployments, minimum baseline should include:

• Auth + headers on protected calls:
  • Authorization: Bearer <token>
  • GSS-Consumer-Id
  • GSS-Consumer-Type
  • GSS-Version
• Token policy:
  • short-lived access tokens (recommended 5-60 minutes)
  • least-privilege scopes per domain/action level
  • optional binding to consumer identity (GSS-Consumer-Id)
• Data guardrails:
  • validate identifiers server-side
  • enforce customer ownership checks before returning payloads
  • fail closed (VALIDATION_ERROR/FORBIDDEN)
• Action controls:
  • request actions with two-step confirmation
  • critical actions with out-of-band verification
• Governance and CI:
  • branch protection on main
  • required checks: lint, tests matrix, coverage, package-check, dependency-audit
  • CODEOWNERS and PR template enabled

Trust Boundary

GSS defines protocol contracts and package logic. Shop implementations own operational security and persistence (token systems, session stores, audit infrastructure).
See docs/compliance-and-trust.md for full guidance.

Testing

pytest

Current test coverage includes:

• happy-path API flows
• forbidden cross-customer order access
• missing header rejection
• protocol enrichment behavior
• CLI integration for login, listing orders, and two-step returns

Documentation

• Spec overview: spec/overview.md
• Architecture deep dive: docs/architecture.md
• API request examples: docs/api-examples.md
• Commands reference: docs/commands-reference.md
• Compliance/trust boundary: docs/compliance-and-trust.md
• Authorization model: docs/authorization-model.md
• Agent delegation model: docs/agent-delegation-model.md
• Multi-language roadmap: docs/multi-language-roadmap.md
• Repository governance: docs/repository-governance.md
• Cloud Run deployment: docs/deploy-cloud-run.md
• Registry security spec: docs/registry-security.md
• Registry conformance checklist: docs/registry-conformance-checklist.md
• Conformance schema: schemas/conformance/agent-delegation-checklist.json
• Shopify webshop project: webshop/shopify-test-store/README.md
• Shop onboarding: docs/getting-started-shops.md
• Consumer onboarding: docs/getting-started-consumers.md
• Protocol format: protocols/FORMAT.md

Contributing

See CONTRIBUTING.md for contribution expectations.

License

This project is licensed under the terms in LICENSE.