Issue Station
Sign In
Tools hereToolbox

openclaw - 💡(How to fix) Fix [Bug]: Telegram channel sessions consistently stall in embedded_run with Codex OAuth, while explicit CLI sessions work

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…

Telegram channel sessions consistently stall in embedded_run when using Codex OAuth, while explicit CLI sessions continue working normally.

Root Cause

Telegram channel sessions consistently stall in embedded_run when using Codex OAuth, while explicit CLI sessions continue working normally.

RAW_BUFFERClick to expand / collapse

Bug type

Regression (worked before, now fails)

Beta release blocker

No

Summary

Telegram channel sessions consistently stall in embedded_run when using Codex OAuth, while explicit CLI sessions continue working normally.

Steps to reproduce

  1. Configure OpenClaw with:

    • Telegram channel enabled
    • openai-codex OAuth auth
    • model: openai/gpt-5.4

  2. Start the gateway: openclaw gateway run

  3. Send a direct Telegram message to the bot (example: "di ok").

  4. Observe logs:

    • Telegram inbound message is received correctly
    • session enters processing/embedded_run
    • no reply is delivered

  5. After ~1-3 minutes, observe repeated stalled session logs such as:

    stalled session: sessionKey=agent:main:telegram:direct:6607485107 classification=stalled_agent_run activeWorkKind=embedded_run lastProgress=codex_app_server:notification:thread/tokenUsage/updated recovery=none

  6. Compare with explicit CLI session, which still succeeds normally:

    openclaw agent --agent main --session-id test-001 --message "say ok"

Expected behavior

A direct Telegram message to the bot should complete the agent run and deliver a reply, as explicit CLI sessions with the same agent/model do:

openclaw agent --agent main --session-id test-001 --message "say ok"

This returns successfully with "ok", so the expected behavior is that the Telegram channel path also completes and sends the response instead of remaining in typing/processing indefinitely.

Actual behavior

Telegram messages are received successfully and the bot enters typing/processing state, but no reply is delivered.

Logs show the session entering embedded_run processing and eventually reporting repeated stalled session diagnostics such as:

stalled session: sessionId=4746794e-b21e-4843-9f9f-40dd73f6c9ef sessionKey=agent:main:telegram:direct:6607485107 state=processing classification=stalled_agent_run activeWorkKind=embedded_run lastProgress=codex_app_server:notification:thread/tokenUsage/updated recovery=none

The Telegram conversation remains stuck in typing state indefinitely.

Meanwhile, explicit CLI sessions continue working normally and return successful responses.

OpenClaw version

2026.5.12 and 2026.5.14-beta.2

Operating system

macOS

Install method

npm global

Model

openai/gpt-5.4

Provider / routing chain

OpenClaw gateway -> codex plugin / embedded_run -> openai-codex OAuth

Additional provider/model setup details

No response

Logs, screenshots, and evidence

Impact and severity

No response

Additional information

No response

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

A direct Telegram message to the bot should complete the agent run and deliver a reply, as explicit CLI sessions with the same agent/model do:

openclaw agent --agent main --session-id test-001 --message "say ok"

This returns successfully with "ok", so the expected behavior is that the Telegram channel path also completes and sends the response instead of remaining in typing/processing indefinitely.

#dependency conflict#environment setup#docker error#permission error#memory optimization

Still need to ship something?

×6

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

Back to top recommendations

TRENDING