[DOCS] Claude Code Default Behaviour Third Party Integration

[shuaibmunshi]

Documentation Type

Missing documentation (feature not documented)

Documentation Location

https://code.claude.com/docs/en/third-party-integrations

Section/Topic

Enterprise deployment overview

Current Documentation

No response

What's Wrong or Missing?

The documentation does not make it clear whether or not Claude Code proxies any data through Anthropic when configured with a third party model provider such as Google Vertex.

I'm guessing a lot of enterprises require specific documentation clearly stating that nothing is being proxied by a tool even when configured to use otherwise approved model providers to avoid obvious privacy concerns.

Suggested Improvement

The documentation should state clearly whether or not Claude Code sends any traffic back to Anthropic when configured to use a model other than any of the Anthropic-provided ones.

Impact

High - Prevents users from using a feature

Additional Context

No response

[github-actions]

---

Found 1 possible duplicate issue:

1. [IMPORTANT] Anthropic Data Gathering with Bedrock #7031

This issue will be automatically closed as a duplicate in 3 days.

• If your issue is a duplicate, please close it and 👍 the existing issue instead
• To prevent auto-closure, add a comment or 👎 this comment

🤖 Generated with Claude Code

[shuaibmunshi]

The comment in that issue answers most of my question but not all. Some enterprise customers need specific verbiage that no data whatsoever is transmitted to Anthropic when Claude Code is configured to use a third party provider. The data usage policy does not explicitly state this fact.

[coygeek]

Issue #16401: Missing Documentation on Data Flow for Third-Party Providers

Status: VALID - Documentation gap for enterprise privacy requirements

Issue URL: #16401

---

Summary

The documentation does not explicitly state whether Claude Code proxies model API traffic through Anthropic when configured with third-party providers (Vertex, Bedrock, Foundry). Enterprises need a clear statement about data flow for compliance and privacy assessments.

---

What Enterprises Need

A clear statement like:

"When configured to use Google Vertex AI, Amazon Bedrock, or Microsoft Foundry, all model API requests go directly to your cloud provider. No prompts, code, or model responses are routed through Anthropic servers."

---

Current Documentation Analysis

What IS Documented

From data-usage.md:79-88 - Default behaviors by API provider:

Service	Claude API	Vertex API	Bedrock API
Statsig (Metrics)	Default on	Default off	Default off
Sentry (Errors)	Default on	Default off	Default off
/bug reports	Default on	Default off	Default off

This shows telemetry is disabled for third-party providers, but doesn't address model API traffic.

From vs-code.md:314:

"By default, Claude Code connects directly to Anthropic's API. If your organization uses Amazon Bedrock, Google Vertex AI, or Microsoft Foundry to access Claude, configure the extension to use your provider instead."

This implies traffic goes to the provider, but doesn't explicitly state "not through Anthropic."

From third-party-integrations.md:116:

"Most organizations can use a cloud provider directly without additional configuration."

Again implies direct connection, but no explicit privacy statement.

What IS NOT Documented

1. Explicit statement that model API calls bypass Anthropic entirely
2. Confirmation that prompts/code never touch Anthropic servers when using third-party providers
3. Clear data flow diagram showing Vertex/Bedrock connections

---

Why This Matters

Enterprises conducting security/compliance reviews need explicit documentation stating:

1. What data flows where - Does prompt text ever reach Anthropic?
2. What's the data path? - Client → Provider only, or Client → Anthropic → Provider?
3. Are there exceptions? - Any scenarios where data still goes to Anthropic?

Without this, legal and security teams cannot approve Claude Code deployment even with approved providers.

---

Recommendation

Add a section to third-party-integrations.md or data-usage.md:

## Data flow with third-party providers

When Claude Code is configured to use Amazon Bedrock, Google Vertex AI, or Microsoft Foundry:

- **Model API requests** (prompts, code, and responses) are sent directly to your
  cloud provider. No model requests are proxied through Anthropic servers.
- **Telemetry and error reporting** are disabled by default (see Default behaviors table).
- **Authentication** is handled by your cloud provider's credentials, not Anthropic.

Your organization's data policies with your cloud provider apply. Anthropic does not
receive or process your prompts when using third-party providers.

---

Files Requiring Update

File	Section	Action
third-party-integrations.md	New section "Data flow with third-party providers"	Add explicit data flow statement
data-usage.md	"Default behaviors by API provider"	Add note clarifying model API traffic goes directly to provider
google-vertex-ai.md	Setup section	Add data flow note
amazon-bedrock.md	Setup section	Add data flow note

[github-actions]

Closing for now — inactive for too long. Please open a new issue if this is still relevant.

[github-actions]

This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.