Bug type

Regression (worked before, now fails)

Beta release blocker

No

Summary

After upgrading from OpenClaw 2026.4.14 to 2026.4.15, several CLI commands became much slower on my machine when Telegram is configured. Downgrading back to 2026.4.14 restored normal speed immediately. This looks like a regression in the Telegram-related config/init/status path in 2026.4.15.

Steps to reproduce

1. Install or upgrade to openclaw@2026.4.15.
2. Configure a working Telegram bot under channels.telegram.
3. Run openclaw status, openclaw doctor --non-interactive, and openclaw channels status.
4. Observe that these commands are significantly slower than expected.
5. Downgrade to openclaw@2026.4.14 and run the same commands again.
6. Observe that command latency returns to normal.

Expected behavior

CLI commands such as openclaw status, openclaw doctor --non-interactive, and openclaw channels status should complete in roughly the same low-latency range as on 2026.4.14, around a few seconds.

Actual behavior

On 2026.4.15, the same commands become much slower when Telegram is configured. In my case:

• openclaw status took about 13 to 15 seconds
• openclaw doctor --non-interactive took about 15 to 16 seconds
• openclaw channels list/status took about 15 to 17 seconds

After downgrading to 2026.4.14, performance returned to normal:

• openclaw status about 4.1 seconds
• openclaw doctor --non-interactive about 4.6 seconds

OpenClaw version

2026.4.15

Operating system

macOS 26.4.1 (arm64)

Install method

npm global

Model

openai-codex/gpt-5.4

Provider / routing chain

openai-codex via local OpenClaw gateway

Additional provider/model setup details

The slowdown does not appear to be model-specific. The issue reproduces on CLI commands like status, doctor, and channels status before normal agent reply generation, and only when Telegram is configured. Downgrading from 2026.4.15 to 2026.4.14 restores normal performance.

Logs, screenshots, and evidence

I benchmarked the affected commands on the same machine before and after downgrade.

Observed on 2026.4.15:

• openclaw status about 13 to 15 seconds
• openclaw doctor --non-interactive about 15 to 16 seconds
• openclaw health --verbose about 11 to 12 seconds
• openclaw channels list/status about 15 to 17 seconds

Observed after downgrade to 2026.4.14:

• openclaw status about 4.1 seconds
• openclaw doctor --non-interactive about 4.6 seconds

Additional isolation:

• minimal config without Telegram was fast
• adding channels.telegram made commands slow again
• channel-specific commands were also slow
• removing large workspace data, old session artifacts, and other local clutter did not materially improve performance

Impact and severity

Affected users: OpenClaw users with Telegram configured on 2026.4.15

Severity: Medium to High, because it significantly slows common diagnostic and operational CLI commands

Frequency: Consistently reproducible on this machine while using 2026.4.15 with Telegram configured

Consequence:

• slower debugging and operations workflow
• slower health/status checks
• degraded local admin experience
• normal performance only returns after downgrade to 2026.4.14

Additional information

Last known good version: 2026.4.14
First known bad version: 2026.4.15

I tested a number of local causes and did not find a material improvement from:

• removing a large workspace backup
• moving a 50GB historical dataset out of the active workspace
• archiving stale session transcript artifacts

The strongest isolation result was that minimal configs were fast, but adding channels.telegram made the slowdown return. Downgrading back to 2026.4.14 restored normal performance immediately.