openclaw - 💡(How to fix) Fix /stop followed by /new silently fails to archive transcript content

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…

Running /stop and then /new in an OpenClaw-managed session silently fails to archive the prior transcript. The new session boots clean but the content of the stopped session is not persisted to the archive — neither to the OpenClaw workspace nor downstream to any ingestion target. The user only finds out hours later when downstream tooling (ingest pipelines, transcript search, retro) reports an empty record for what was clearly a working session.

Error Message

  • /stop returns success status.
  • /new returns success status with a new session ID.
  • The archive contains no entry for the stopped session (or an empty stub).
  • Transcript content from the stopped session is lost.

Root Cause

This breaks downstream ingestion (gbrain claude-sessions / telegram-sessions pipelines) because the archive is the ingestion surface. Lost transcripts = lost knowledge-brain coverage with no obvious signal at the time of loss.

RAW_BUFFERClick to expand / collapse

Summary

Running /stop and then /new in an OpenClaw-managed session silently fails to archive the prior transcript. The new session boots clean but the content of the stopped session is not persisted to the archive — neither to the OpenClaw workspace nor downstream to any ingestion target. The user only finds out hours later when downstream tooling (ingest pipelines, transcript search, retro) reports an empty record for what was clearly a working session.

Environment

  • OpenClaw version: 2026.5.7 (installed from npm)
  • Host: Ubuntu 24.04, openclaw-gateway running as systemd user unit
  • ExecStart: /usr/bin/node /home/madair/.npm-global/lib/node_modules/openclaw/dist/index.js gateway --port 18789
  • Telegram bot path active

Reproduction

  1. Start an OpenClaw session via the gateway.
  2. Run commands, generate transcript content (>1 turn).
  3. Invoke /stop to end the current session.
  4. Immediately invoke /new to start a fresh session.
  5. Check the archive location (~/.openclaw/workspace/... or configured archive path) for the prior session's transcript.

Observed behavior

  • /stop returns success status.
  • /new returns success status with a new session ID.
  • The archive contains no entry for the stopped session (or an empty stub).
  • Transcript content from the stopped session is lost.

Expected behavior

  • /stop should atomically flush the in-memory transcript to the archive before returning success.
  • /new should not be reachable until the prior /stop's archive write is durable.
  • An archive-failure should surface as a non-zero exit / error toast, not as silent success.

Cost shape

This breaks downstream ingestion (gbrain claude-sessions / telegram-sessions pipelines) because the archive is the ingestion surface. Lost transcripts = lost knowledge-brain coverage with no obvious signal at the time of loss.

Suggested fix direction

  • Either make /stop synchronous on the archive write (fail-loud if write fails), or have /new block on the prior session's archive completion.
  • Emit a structured journal line session.archived: { session_id, byte_count } on successful archive so downstream consumers can verify.

Reference

Surfaced in the 2026-05-13 PowerStack system-health audit (ticket ADA-442, Phase 3.6). PowerStack run notes attached to the audit page in our local knowledge brain.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…

FAQ

Expected behavior

  • /stop should atomically flush the in-memory transcript to the archive before returning success.
  • /new should not be reachable until the prior /stop's archive write is durable.
  • An archive-failure should surface as a non-zero exit / error toast, not as silent success.

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

openclaw - 💡(How to fix) Fix /stop followed by /new silently fails to archive transcript content