shareAI-lab/learn-claude-code · error · RuntimeError

unable to load integrated host from {path}

Error message

unable to load integrated host from {path}

What it means

load_integrated_host() lazily imports the s15 harness by building an importlib spec from the sibling path s15_integrated_harness/code.py; if spec or spec.loader is None (which importlib returns for paths it cannot load, e.g. missing file or unrecognized extension) it raises RuntimeError. This is a host-assembly/environment error, not a workflow-input error — it means the bundled demo/test harness files are not where the module expects them.

Source

Thrown at s16_workflow_runtime/code.py:802

    base_assemble = host.assemble_tool_pool

    def assemble_with_workflow():
        tools, handlers = base_assemble()
        if not any(tool.get("name") == "Workflow" for tool in tools):
            tools.append(WORKFLOW_TOOL)
        handlers["Workflow"] = run_workflow_sync
        return tools, handlers

    host.assemble_tool_pool = assemble_with_workflow
    host._workflow_tool_installed = True


def load_integrated_host():
    """Load s15 lazily so deterministic workflow tests need no API key."""
    path = Path(__file__).resolve().parents[1] / "s15_integrated_harness" / "code.py"
    spec = importlib.util.spec_from_file_location("integrated_host", path)
    if spec is None or spec.loader is None:
        raise RuntimeError(f"unable to load integrated host from {path}")
    host = importlib.util.module_from_spec(spec)
    sys.modules[spec.name] = host
    spec.loader.exec_module(host)
    return host


# -- CLI --
async def run_demo(argv):
    resume_id = None
    if argv and argv[0] == "resume":
        resume_id = _read_last_run()
        if not resume_id:
            print("nothing to resume; run `python code.py demo` first.")
            return
        print(f"resuming {resume_id}; unchanged agent() calls use the journal cache\n")
    else:
        print("launching workflow `review-changes`\n")

View on GitHub (pinned to 985456f4ad)

Solutions

1. Restore the sibling directory so <repo>/s15_integrated_harness/code.py exists next to s16_workflow_runtime.
2. If the layout changed, update the hardcoded path in load_integrated_host() to the new location.
3. Skip features that need the integrated host (deterministic tests requiring no API key) or vendor the s15 module into the deployment.

Example fix

# before
path = Path(__file__).resolve().parents[1] / "s15_integrated_harness" / "code.py"
# missing in this deployment -> RuntimeError

# after
path = Path(__file__).resolve().parents[1] / "harness" / "s15_code.py"  # actual location
assert path.exists(), f"integrated host missing at {path}"

from pathlib import Path
host_path = Path(__file__).resolve().parents[1] / "s15_integrated_harness" / "code.py"
if not host_path.is_file():
    raise FileNotFoundError(f"s15 harness missing at {host_path}; cannot assemble integrated host")

try:
    host = load_integrated_host()
except RuntimeError as e:
    if "unable to load integrated host" in str(e):
        log_env_issue(e)  # report packaging/checkout problem; degrade to non-integrated mode
    raise

When it happens

Trigger: Calling load_integrated_host() when s15_integrated_harness/code.py is absent from the repo checkout, or packaging that ships s16 without its sibling directory so spec_from_file_location yields no usable loader.

Common situations: Partial clones or trimmed deployment artifacts that omit s15; running s16 standalone from an installed location instead of the repository tree; renames/moves of the s15 directory breaking the parents[1] relative path.

Related errors

• Unable to load memory runtime from {path}
• Not in a git repository. worktree tools require git.
• Memory directory escapes the workspace
• could not allocate a unique workflow runId
• invalid workflow runId

AI-assisted analysis of shareAI-lab/learn-claude-code@985456f4ad (2026-08-14). Data as JSON: /api/errors/8f2dfddb89c17d00. Report an issue: GitHub.