Cairn

Stale failure-turn requests executing in later messages

NousResearch/hermes-agent · cluster 107070

Data as of

Nothing open on GitHub. Fix merged in #108033. Check it covers your case.

The short answer

0open issues
0open PRs
6items grouped
Known?
Yes. 6 related items grouped here, 1 issue already closed.
Fix event
A fix was merged to main in PR #108033. Check it covers your case. No release data is tracked.
Tracker
The maintainers' tracker marks this resolved, deciding PR #108033. That is separate from the fix event.
Still open
Nothing open on GitHub.

These issues and fixes address a bug where requests from incomplete or erroring turns were persisted and later replayed into subsequent messages, causing stale or unintended mutations to execute. The work focuses on preventing failed turns from leaking into future message processing.

Open issues0

No open issues.

Open PRs0

No open pull requests.

Fixes on record

Merged to main. A stamp counts the issues a fix is linked to.

  • #108033 fix(gateway): a failed turn's request can no longer ride into the next message and execute later (#107070, salvage #107088)

    merged to main

    files: gateway/run_turn.py, gateway/session_transcript.py, hermes_state_messages.py, tests/gateway/test_42039_duplicate_user_message.py, tests/gateway/test_failure_writer_ownership.py, tests/gateway/test_incomplete_gateway_turns.py

4 PRs closed without merging

From the close comments: duplicate 1 · merged via another change 1 · partial fix 1 · superseded by another PR 1

  • #107088 fix(gateway): stop stale failed requests from executing later · partial fix → 32570e9
  • #107108 fix(agent): age-gate stale unanswered user merges in sequence repair · merged via another change → 044a77b, #107070, #108033, #107088
  • #107561 fix(gateway): make failed-turn persistence exactly-once · duplicate → #107088, #107070, #107438, #107108
  • #107980 fix(gateway): a failed turn's request can no longer ride into the next message and execute later (#107070, salvage #107088) · superseded by another PR
Closed issues · 1
  • Closed#107070[Bug]: user turns persisted after a transient failure are replayed days later — alternation repair merges them into the next message and the model executes stale mutating requests
How this cluster was grouped
  • Link evidence (structural links throughout): every member is connected by structural evidence: shared changed files, or a closing/superseded reference.
  • Pairwise link density 0.33: the share of item pairs that carry a link, counting a structural link in full and a relevance-only link at 0.36. 1.00 means every pair is linked.

Pick up a Lab task

Check whether a fix covers the open issues here, with your own agent. Bring back evidence for maintainers.

Open the Lab

At a glance

Merged PRs
#108033
Closed issues
1
Attempts
4 PRs closed without merging
Fix link checked
12 Sep 2026

For your agent

The same facts, typed.

cluster/107070.json
cluster/107070.md