Add custom headers from the env var LLM_EXTRA_HEADERS

docs/advanced/configuration.mdx

@@ -35,6 +35,11 @@ Configure Strix using environment variables or a config file.
   Timeout in seconds for memory compression operations (context summarization).
 </ParamField>
 
+<ParamField path="LLM_EXTRA_HEADERS" type="string">
+  Custom HTTP headers to include in every LiteLLM request. Accepts a JSON object,
+  e.g. `{"x-my-header": "value"}`.
+</ParamField>
+
 ## Optional Features
 
 <ParamField path="PERPLEXITY_API_KEY" type="string">

strix/config/config.py

@@ -18,6 +18,7 @@ class Config:
     openai_api_base = None
     litellm_base_url = None
     ollama_api_base = None
+    llm_extra_headers = None
     strix_reasoning_effort = "high"
     strix_llm_max_retries = "5"
     strix_memory_compressor_timeout = "30"
@@ -29,6 +30,7 @@ class Config:
         "openai_api_base",
         "litellm_base_url",
         "ollama_api_base",
+        "llm_extra_headers",
         "strix_reasoning_effort",
         "strix_llm_max_retries",
         "strix_memory_compressor_timeout",
@@ -196,18 +198,20 @@ def save_current_config() -> bool:
     return Config.save_current()
 
 
-def resolve_llm_config() -> tuple[str | None, str | None, str | None]:
-    """Resolve LLM model, api_key, and api_base based on STRIX_LLM prefix.
+def resolve_llm_config() -> tuple[str | None, str | None, str | None, dict[str, str] | None]:
+    """Resolve LLM model, api_key, api_base based on STRIX_LLM prefix
+        and extra_headers for LiteLLM calls.
 
     Returns:
-        tuple: (model_name, api_key, api_base)
+        tuple: (model_name, api_key, api_base, extra_headers)
         - model_name: Original model name (strix/ prefix preserved for display)
         - api_key: LLM API key
         - api_base: API base URL (auto-set to STRIX_API_BASE for strix/ models)
+        - extra_headers : Custom headers
     """
     model = Config.get("strix_llm")
     if not model:
-        return None, None, None
+        return None, None, None, None
 
     api_key = Config.get("llm_api_key")
 
@@ -221,4 +225,16 @@ def resolve_llm_config() -> tuple[str | None, str | None, str | None]:
             or Config.get("ollama_api_base")
         )
 
-    return model, api_key, api_base
+    extra_headers: dict[str, str] = {}
+    raw_headers = Config.get("llm_extra_headers") or ""
+    if raw_headers.strip():
+        try:
+            parsed = json.loads(raw_headers)
+            if isinstance(parsed, dict):
+                extra_headers = {str(k): str(v) for k, v in parsed.items() if v is not None}
+            else:
+                raise TypeError("LLM_EXTRA_HEADERS must be a JSON object")
+        except json.JSONDecodeError as e:
+            raise ValueError(f"Invalid LLM_EXTRA_HEADERS JSON: {e}") from e
+
+    return model, api_key, api_base, extra_headers

strix/interface/main.py

@@ -5,6 +5,7 @@
 
 import argparse
 import asyncio
+import json
 import logging
 import os
 import shutil
@@ -82,6 +83,36 @@ def validate_environment() -> None:  # noqa: PLR0912, PLR0915
     if not Config.get("strix_reasoning_effort"):
         missing_optional_vars.append("STRIX_REASONING_EFFORT")
 
+    raw_headers = Config.get("llm_extra_headers") or ""
+    if raw_headers.strip():
+        try:
+            parsed = json.loads(raw_headers)
+            if not isinstance(parsed, dict):
+                raise TypeError("LLM_EXTRA_HEADERS must be a JSON object, got a non-dict value")
+        except (json.JSONDecodeError, ValueError) as e:
+            error_text = Text()
+            error_text.append("INVALID LLM_EXTRA_HEADERS", style="bold red")
+            error_text.append("\n\n", style="white")
+            error_text.append("LLM_EXTRA_HEADERS must be a valid JSON object.\n", style="white")
+            error_text.append(f"Error: {e}\n", style="white")
+            error_text.append("\nExample:\n", style="white")
+            error_text.append(
+                'export LLM_EXTRA_HEADERS={"x-my-header": "value"}\n',
+                style="dim white",
+            )
+
+            panel = Panel(
+                error_text,
+                title="[bold white]STRIX",
+                title_align="left",
+                border_style="red",
+                padding=(1, 2),
+            )
+            console.print("\n")
+            console.print(panel)
+            console.print()
+            sys.exit(1)
+
     if missing_required_vars:
         error_text = Text()
         error_text.append("MISSING REQUIRED ENVIRONMENT VARIABLES", style="bold red")
@@ -208,7 +239,7 @@ async def warm_up_llm() -> None:
     console = Console()
 
     try:
-        model_name, api_key, api_base = resolve_llm_config()
+        model_name, api_key, api_base, extra_headers = resolve_llm_config()
         litellm_model, _ = resolve_strix_model(model_name)
         litellm_model = litellm_model or model_name
 
@@ -228,6 +259,8 @@ async def warm_up_llm() -> None:
             completion_kwargs["api_key"] = api_key
         if api_base:
             completion_kwargs["api_base"] = api_base
+        if extra_headers:
+            completion_kwargs["extra_headers"] = extra_headers
 
         response = litellm.completion(**completion_kwargs)
 

strix/llm/config.py

@@ -18,7 +18,7 @@ def __init__(
         reasoning_effort: str | None = None,
         system_prompt_context: dict[str, Any] | None = None,
     ):
-        resolved_model, self.api_key, self.api_base = resolve_llm_config()
+        resolved_model, self.api_key, self.api_base, self.extra_headers = resolve_llm_config()
         self.model_name = model_name or resolved_model
 
         if not self.model_name:

strix/llm/dedupe.py

@@ -156,7 +156,7 @@ def check_duplicate(
 
         comparison_data = {"candidate": candidate_cleaned, "existing_reports": existing_cleaned}
 
-        model_name, api_key, api_base = resolve_llm_config()
+        model_name, api_key, api_base, extra_headers = resolve_llm_config()
         litellm_model, _ = resolve_strix_model(model_name)
         litellm_model = litellm_model or model_name
 
@@ -181,6 +181,8 @@ def check_duplicate(
             completion_kwargs["api_key"] = api_key
         if api_base:
             completion_kwargs["api_base"] = api_base
+        if extra_headers:
+            completion_kwargs["extra_headers"] = extra_headers
 
         response = litellm.completion(**completion_kwargs)
 

strix/llm/llm.py

@@ -286,6 +286,8 @@ def _build_completion_args(self, messages: list[dict[str, Any]]) -> dict[str, An
             args["api_key"] = self.config.api_key
         if self.config.api_base:
             args["api_base"] = self.config.api_base
+        if self.config.extra_headers:
+            args["extra_headers"] = self.config.extra_headers
         if self._supports_reasoning():
             args["reasoning_effort"] = self._reasoning_effort
 

strix/llm/memory_compressor.py

@@ -104,7 +104,7 @@ def _summarize_messages(
     conversation = "\n".join(formatted)
     prompt = SUMMARY_PROMPT_TEMPLATE.format(conversation=conversation)
 
-    _, api_key, api_base = resolve_llm_config()
+    _, api_key, api_base, extra_headers = resolve_llm_config()
 
     try:
         completion_args: dict[str, Any] = {
@@ -116,6 +116,8 @@ def _summarize_messages(
             completion_args["api_key"] = api_key
         if api_base:
             completion_args["api_base"] = api_base
+        if extra_headers:
+            completion_args["extra_headers"] = extra_headers
 
         response = litellm.completion(**completion_args)
         summary = response.choices[0].message.content or ""