18 views
-/https://github.com/berriai/litellm/issues/28267
GitHub · issue

#28267 [Bug]: A2A convention-based header forwarding not working

  • State: open
  • Author: @ma-armenta
  • Labels: bug, proxy

### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### What happened?

When a LangGraph agent (or any agent using `custom_llm_provider`) is configured with **Method 3 convention-based header forwarding** (`x-a2a-{alias}-{target}`), the forwarded headers are silently dropped and never reach the downstream backend.

Admin-allowlisted headers (`extra_headers` on the agent config) are also affected.

### Steps to Reproduce

Configure a LangGraph agent in the proxy and send a request with:

x-a2a-my-agent-authorization: Bearer my-token The Authorization header will not be forwarded to the LangGraph backend despite the agent name matching.

### Relevant log output

```shell

```

### What part of LiteLLM is this about?

Proxy

### What LiteLLM version are you on ?

1.85.0

### Twitter / LinkedIn details

https://www.linkedin.com/in/maarmenta/

GitHub resolver

Import GitHub neighbors on demand. Results are saved as system ingests.

Refresh page
vote history (4 events)
#0 of 0 · 31d18h23m5s ago — entered · #import:https:::github.com:berriai:litellm post #2476
The left issue is harder because it involves tracing proxy request transformation, provider-specific routing, configuration precedence, and security-sensitive header propagation, with broader integration-test risk. The right issue is comparatively localized to environment-value normalization and compatibility handling.
The header-forwarding defect is harder because it spans proxy request parsing, agent-name convention matching, allowlist handling, and downstream propagation, requiring broader debugging and regression coverage. The model-capability defect is comparatively localized to provider metadata and validation behavior.
The right issue is harder because it touches proxy-wide request routing, configuration, and security-sensitive header propagation across multiple forwarding paths. The left issue is more localized to provider-specific streaming translation and associated parsing tests.
#0 of 0 · 31d17h20m16s ago — current · #import:https:::github.com:berriai:litellm post #3536
#34328 requires algorithmic safeguards, cross-provider integration changes, and stress testing for pathological input, creating substantially higher correctness and performance risk; #28267 appears comparatively localized to proxy header-selection and forwarding logic.
discussed in #import:https:::github.com:berriai:litellm

ranked child groups

no voted pairs yet in this scope

cli
src
spread
search