Add Dynamic/Runtime Header by Langchain RunnableConfig

[jpfcabral]

Fixes: #280 #312 #271 #235 #233 #194 #177

Added:

• Client header injection by RunnableConfig

Example:

await graph.ainvoke(input, config={"configurable": {"jwt": "abc"}})

await graph.ainvoke(input, config={"configurable": {"mcp_headers": {"x-functions-key": "abc"}}}

await tool.ainvoke(input, config={"configurable": {"jwt": "abc"}})

await tool.ainvoke(input, config={"configurable": {"mcp_headers": {"x-functions-key": "abc"}}}

It should work for any langchain runnable (chains, chatmodels, tools, etc)

Reference:
https://python.langchain.com/api_reference/core/runnables/langchain_core.runnables.config.RunnableConfig.html
https://python.langchain.com/docs/how_to/tool_configure/

[eyurtsev]

Hi @jpfcabral thanks for the PR. This design is not configurable on a per tool basis.

We're likely going to this using an interceptor pattern

[jpfcabral]

Awesome, @eyurtsev !!! interceptor pattern will be better as #313 implementation only solves the problem of dynamic headers, which is blocking production scenarios for app based services!

Hoping final solution soon!! 😃 tks

[jzus19]

@eyurtsev when to expect solution?