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

Unable to load memory runtime from {path}

Error message

Unable to load memory runtime from {path}

What it means

load_memory_runtime() dynamically imports the s09_memory module via importlib.util.spec_from_file_location; if the returned spec or its loader is None, Python could not build an importable module for the file at <repo>/s09_memory/code.py. This is a host-environment failure at harness startup (the call runs at module import time), not a runtime argument error.

Source

Thrown at s15_integrated_harness/code.py:85

MAX_CONSECUTIVE_529 = 2
MAX_RECOVERY_RETRIES = 2
BASE_DELAY_MS = 500
CONTEXT_LIMIT = 50000
KEEP_RECENT_TOOL_RESULTS = 3
PERSIST_THRESHOLD = 30000
CONTINUATION_PROMPT = "Continue from the previous response. Do not repeat completed work."
PROMPT = "\033[36ms15 >> \033[0m"
CLI_ACTIVE = False


def load_memory_runtime():
    """Load s09 once and share this host's client, model, and workspace."""
    path = Path(__file__).resolve().parents[1] / "s09_memory" / "code.py"
    spec = importlib.util.spec_from_file_location(
        f"integrated_memory_{id(client)}", path
    )
    if spec is None or spec.loader is None:
        raise RuntimeError(f"Unable to load memory runtime from {path}")
    runtime = importlib.util.module_from_spec(spec)
    spec.loader.exec_module(runtime)
    runtime.WORKDIR = WORKDIR
    runtime.MEMORY_DIR = WORKDIR / ".memory"
    runtime.MEMORY_INDEX = runtime.MEMORY_DIR / "MEMORY.md"
    runtime.client = client
    runtime.MODEL = MODEL
    return runtime


MEMORY_RUNTIME = load_memory_runtime()


class ConsoleBroker:
    """Serialize normal prompts and worker permission questions on one stdin."""

    def __init__(self):
        self._lock = threading.Lock()

View on GitHub (pinned to 985456f4ad)

Solutions

1. Restore or copy the s09_memory/code.py module so it sits next to s15_integrated_harness (parent-relative path: <repo>/s09_memory/code.py).
2. If you relocated s09, patch the path in load_memory_runtime() to point at the new location.
3. Check file readability (permissions, mount) for s09_memory/code.py before re-running the harness.

from pathlib import Path

path = Path(__file__).resolve().parents[1] / "s09_memory" / "code.py"
if not path.is_file():
    raise SystemExit(f"Missing dependency: {path}. Clone/place s09_memory next to s15_integrated_harness.")

try:
    MEMORY_RUNTIME = load_memory_runtime()
except RuntimeError as e:
    raise SystemExit(f"startup failed: {e}") from e

When it happens

Trigger: Running s15_integrated_harness/code.py when the sibling directory s09_memory/code.py is missing, is a directory, has an unsupported/unknown extension (no .py suffix handled), or when the file cannot be stat'd due to permissions. spec_from_file_location returns None for unrecognized file types; loader can be None for namespace-package-like paths.

Common situations: Copying s15 without its s09 dependency sibling; partial clone or archive that skipped s09_memory; renaming code.py (e.g. to code.pyx or memory.py) without updating the hardcoded relative path; read-only mount where the path resolves but cannot be loaded.

Related errors

• unable to load integrated host from {path}
• Not in a git repository. worktree tools require git.
• Memory directory escapes the workspace
• Install dependencies first: pip install -r requirements.txt
• MODEL_ID is required in the environment or .env

AI-assisted analysis of shareAI-lab/learn-claude-code@985456f4ad (2026-08-14). Data as JSON: /api/errors/9bce882399ab7ee5. Report an issue: GitHub.