model-providers

model-providers with GTM OS

When to use this skill

Use when selecting or configuring Anthropic, OpenAI, or Gemini as a GTM OS Model Gateway provider for reasoning or embeddings.

What GTM OS uses it for

Native ModelBackend providers: Anthropic generate, OpenAI generate and embeddings, Gemini generate and embeddings. The gateway chooses a provider using model policy routes.

Capabilities and operation names are listed below as they appear in the connector package and coverage map. Discover the live input and output schema with gtm describe before invocation; the registry for the current organization is authoritative if it differs from this static guide.

• Providers register with the ModelProviderRegistry; do not invoke provider SDKs directly from agent workflow code.
• Anthropic is generate-only; OpenAI (text-embedding-3-small) and Gemini (gemini-embedding-001) serve generation and dimensions-aware embeddings -- whichever key the org has, or a definition.model_policy route for the embedding task class.
• Missing key degrades to the kernel fixture backend. Region-specific routing fails closed without a configured endpoint.

Connect it

Configure org credentials through the model provider connection/credential flow, or operator env ANTHROPIC_API_KEY, OPENAI_API_KEY, GEMINI_API_KEY, or OPENROUTER_API_KEY (OpenAI-compatible, many vendors). Select a provider/model in model policy; the package source shows supported route keys. .env.example documents Anthropic/OpenAI; coverage documents Gemini.

These providers use API keys, not OAuth scopes in the GTM OS adapter. Use separate least-privilege project keys and keep them in sealed connections or server env. Region routing requires a matching configured regional endpoint or attestation; do not silently fall back across regions.

Use the console Connections/settings page for the provider when present. For CLI setup, inspect gtm connect --help and use the provider name supported by the installed CLI; do not guess flags or paste credentials into command history. Restart/reload the configured instance only through its documented operator workflow.

Canonical data

• Model calls return generated text or embeddings plus usage, cost, and provenance to the gateway.
• They do not directly read/write gtm.* records; the invoking operation determines object access.
• OpenAI and Gemini embeddings must respect requested dimensions.

Read and write records through registered operations and gtm.* objects only. A vendor-side record is not automatically a canonical GTM OS record; use the connector's declared mapping and retain source provenance. Do not infer fields from the vendor API.

Governance and safety

Model spend is quoted/reserved/settled by the gateway; respect route budgets and token limits. Follow data privacy/ZDR policy: OpenAI uses request-level store:false; Anthropic and Gemini ZDR is operator/account attestation, not a per-request wire flag. Do not send suppressed personal data into prompts without authorization.

• E-class/customer-facing writes are proposed for approval; a different human approves them.
• Respect consent, suppression, unsubscribe, and do-not-contact state before any person-level action.
• Honor configured spend/budget caps. Do not retry a billable request blindly after timeout.
• Keep connection secrets in the sealed connection or server environment. Never print or commit them.

Common errors and recovery

• No API key: fixture backend may answer; don't call it live provider output.

• Region unavailable: configure the correct regional endpoint; fail closed.

• 429/transport error: gateway may fail over according to policy; don't create a parallel raw SDK call.

• Wrong capability: embeddings come from OpenAI or Gemini; Anthropic has no embedding model (generate only). No OpenAI/Gemini key: semantic search is text-only.

• Check the connection state and provider health before retrying; use the connector's fixture mode only for development.

• Read the operation error and lineage/receipt before changing inputs or repeating a write.

• If the provider is not registered, stop and report that limitation; do not call the vendor directly as a workaround.

Do

• Read the connector entry in docs/gtmos/reference/connections.md and the current gtm describe schema before a new integration action.
• Start with a read or one narrowly scoped test, confirm provenance, then scale within the approved budget.
• Ask a human to resolve missing consent, expired credentials, or approval-required actions.

Don't

• Don't invent endpoints, scopes, operation names, or object fields.
• Don't bypass GTM OS governance with direct vendor writes or direct database edits.
• Don't send messages, change campaigns, or publish content without the operation's approval path.

Official vendor documentation

Anthropic: https://docs.anthropic.com/ ; OpenAI: https://platform.openai.com/docs/ ; Gemini: https://ai.google.dev/gemini-api/docs

GTM OS reference: docs/gtmos/reference/connections.md (installed by gtm init) and the live gtm describe schema.