usestrix/strix · critical · RuntimeError

No LLM model configured. Set STRIX_LLM env or pass model= to

Error message

No LLM model configured. Set STRIX_LLM env or pass model= to run_strix_scan().

What it means

Raised at the top of run_strix_scan (strix/core/runner.py:173) when no LLM model can be resolved: the model= argument, the loaded settings' llm.model (typically set from the STRIX_LLM environment variable), are all empty after stripping. Strix is BYO-LLM-key and needs a LiteLLM model id for every agent turn, so it fails fast with RuntimeError before creating any sandbox session.

Source

Thrown at strix/core/runner.py:173

    agents_path = state_dir / "agents.json"
    agents_db = state_dir / "agents.db"
    is_resume = agents_path.exists()

    logger.info(
        "%s Strix scan %s (image=%s, max_turns=%d, interactive=%s, run_dir=%s)",
        "Resuming" if is_resume else "Starting",
        scan_id,
        image,
        max_turns,
        interactive,
        run_dir,
    )

    settings = load_settings()
    configure_sdk_model_defaults(settings)
    resolved_model = (model or settings.llm.model or "").strip()
    if not resolved_model:
        raise RuntimeError(
            "No LLM model configured. Set STRIX_LLM env or pass model= to run_strix_scan().",
        )
    logger.info("LLM model resolved: %s", resolved_model)
    chat_completions_tools = uses_chat_completions_tool_schema(resolved_model, settings)

    if coordinator is None:
        coordinator = AgentCoordinator()
    coordinator.set_snapshot_path(agents_path)

    from strix.tools.notes.tools import hydrate_notes_from_disk
    from strix.tools.todo.tools import hydrate_todos_from_disk

    hydrate_todos_from_disk(state_dir)
    hydrate_notes_from_disk(state_dir)

    root_id: str | None = None
    if is_resume:
        try:

View on GitHub (pinned to 8551339130)

Solutions

1. export STRIX_LLM="<litellm-model-id>" (e.g. openai/gpt-4o, anthropic/claude-sonnet-4) in the shell/profile that launches strix.
2. Or pass model= explicitly when calling run_strix_scan() programmatically.
3. Or persist it in ~/.strix/cli-config.json under llm.model so shells without the env var still work.
4. Verify with `strix config` (or load_settings()) that llm.model is non-empty, and confirm LLM_API_KEY is also set for that provider.

Example fix

# before
$ strix -n -t ./  # RuntimeError: No LLM model configured

# after
$ export STRIX_LLM="openai/gpt-4o"
$ export LLM_API_KEY="sk-..."
$ strix -n -t ./

import os
from strix.config import load_settings

model = (os.environ.get('STRIX_LLM') or load_settings().llm.model or '').strip()
if not model:
    raise SystemExit('Set STRIX_LLM or llm.model before starting a scan')

def has_llm_model() -> bool:
    import os
    from strix.config import load_settings
    return bool((os.environ.get('STRIX_LLM') or load_settings().llm.model or '').strip())

try:
    await run_strix_scan(target=..., model=None)
except RuntimeError as exc:
    if 'No LLM model configured' in str(exc):
        os.environ['STRIX_LLM'] = 'openai/gpt-4o'
        await run_strix_scan(target=..., model=None)
    else:
        raise

When it happens

Trigger: Calling run_strix_scan() without model= while STRIX_LLM is unset/empty and ~/.strix/cli-config.json has no llm.model; running the CLI in a fresh shell where the env var was never exported; CI jobs that strip the environment.

Common situations: Fresh installs that never ran `strix config`; env vars lost through sudo, cron, Docker, or systemd service units; typos like STRIX_LLM="" or exporting in a subshell.

Related errors

• No model configured. Set STRIX_LLM first.
• extra_system_prompt_context cannot override built-in scope k
• Model connection failed: {exc}
• strix_image is not configured. Set it in ~/.strix/cli-config
• The interactive interface could not start: {exc}

AI-assisted analysis of usestrix/strix@8551339130 (2026-08-15). Data as JSON: /api/errors/336b26d0c42a830e. Report an issue: GitHub.