fix(vertex_ai): support pluggable (executable) credential_source for WIF auth

[michelligabriele]

The WIF credential dispatch in load_auth() only handled identity_pool and aws credential types. When credential_source.executable was present (used for Azure Managed Identity via Workload Identity Federation), it fell through to identity_pool.Credentials which rejected it with MalformedError.

Add dispatch to google.auth.pluggable.Credentials for executable-type credential sources, following the same pattern as the existing identity_pool and aws helpers.

Fixes authentication for Azure Container Apps → GCP Vertex AI via WIF with executable credential sources.

Relevant issues

Pre-Submission checklist

Please complete all items before asking a LiteLLM maintainer to review your PR

• I have Added testing in the tests/test_litellm/ directory, Adding at least 1 test is a hard requirement - see details
• My PR passes all unit tests on make test-unit
• My PR's scope is as isolated as possible, it only solves 1 specific problem
• I have requested a Greptile review by commenting @greptileai and received a Confidence Score of at least 4/5 before requesting a maintainer review

Delays in PR merge?

If you're seeing a delay in your PR being merged, ping the LiteLLM Team on Slack (#pr-review).

CI (LiteLLM team)

CI status guideline:

• 50-55 passing tests: main is stable with minor issues.
• 45-49 passing tests: acceptable but needs attention
• <= 40 passing tests: unstable; be careful with your merges and assess the risk.

• Branch creation CI run
  Link:

• CI run for the last commit
  Link:

• Merge / cherry-pick CI run
  Links:

Type

🐛 Bug Fix

Changes

• litellm/llms/vertex_ai/vertex_llm_base.py: Added elif branch in the WIF dispatch logic (inside load_auth()) to check for credential_source.executable and route to new _credentials_from_pluggable() helper, which calls google.auth.pluggable.Credentials.from_info(). Follows the same pattern as the existing _credentials_from_identity_pool() and _credentials_from_identity_pool_with_aws() helpers.
• tests/test_litellm/llms/vertex_ai/test_vertex_llm_base.py: Added 3 unit tests:
  • test_credentials_from_pluggable_implementation — verifies helper dispatches to pluggable.Credentials with scopes
  • test_credentials_from_pluggable_no_scopes_needed — verifies helper works when scopes not required
  • test_load_auth_dispatches_to_pluggable_for_executable — verifies load_auth() routes executable credential_source to pluggable (not identity_pool)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
litellm	Ready	Preview, Comment	Mar 27, 2026 8:38pm

[greptile-apps]

Greptile Summary

This PR fixes a MalformedError crash that occurred when WIF credentials used an executable-type credential_source (e.g. Azure Managed Identity → GCP Vertex AI via Workload Identity Federation). The load_auth() dispatch logic previously only handled aws environment IDs and fell through to identity_pool.Credentials, which rejected executable sources. The fix adds a targeted elif branch and a new _credentials_from_pluggable() helper that mirrors the pattern of the existing _credentials_from_identity_pool() and _credentials_from_identity_pool_with_aws() helpers.\n\n- Added elif isinstance(credential_source, dict) and \"executable\" in credential_source branch in load_auth() to route pluggable credentials correctly.\n- Added _credentials_from_pluggable() using google.auth.pluggable.Credentials.from_info(), consistent with the surrounding helpers (same import-guard, same requires_scopes check, same scopes application).\n- Added 3 unit tests covering: scoped dispatch, no-scopes-needed path, and the load_auth() routing — all fully mocked with no real network calls, satisfying the test-folder policy.

Confidence Score: 5/5

Safe to merge — minimal, well-scoped fix that adds a missing dispatch branch without touching any existing code paths.

The change is a pure additive elif branch and a new helper method; no existing behavior is altered. The new helper is a direct copy of the established _credentials_from_identity_pool pattern (same import guard, same requires_scopes check). Three new mock tests cover scoped, unscoped, and routing scenarios with no real network calls. No P0 or P1 findings identified.

No files require special attention.

Important Files Changed

Filename	Overview
litellm/llms/vertex_ai/vertex_llm_base.py	Adds elif dispatch branch for executable credential sources and a new _credentials_from_pluggable() helper, following the exact same pattern as existing identity_pool and AWS helpers.
tests/test_litellm/llms/vertex_ai/test_vertex_llm_base.py	Adds 3 new mock-only tests covering the pluggable helper and the load_auth() dispatch; no existing tests modified, no real network calls.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[load_auth called with external_account credentials] --> B{credential_source\nenvironment_id contains 'aws'?}
    B -->|Yes| C{Explicit AWS params present?}
    C -->|Yes| D[VertexAIAwsWifAuth.credentials_from_explicit_aws]
    C -->|No| E[_credentials_from_identity_pool_with_aws]
    B -->|No| F{credential_source\nhas 'executable' key?}
    F -->|Yes| G[_credentials_from_pluggable\ngoogle.auth.pluggable.Credentials]
    F -->|No| H[_credentials_from_identity_pool\ngoogle.auth.identity_pool.Credentials]
    G --> I[Apply scopes if requires_scopes]
    H --> I
    D --> I
    E --> I
    I --> J[refresh_auth & return creds, project_id]

Loading

_{Reviews (1): Last reviewed commit: "fix(vertex_ai): support pluggable (execu..." | Re-trigger Greptile}

[codspeed-hq]

Merging this PR will not alter performance

✅ 16 untouched benchmarks

---

_{Comparing michelligabriele:fix/vertex-wif-pluggable-credential-dispatch (1704bd0) with main (d949085)}