Telegram bridge failing with "Network request failed" in NVIDIA Sandbox environment

[disolutionslt]

Description:
I am trying to run NemoClaw in the NVIDIA sandbox environment, but the Telegram channel fails to initialize. Even after fixing the PATH for openshell and ensuring the botToken is correct, the gateway keeps throwing network errors.

Error logs:

Plaintext
[telegram] telegram deleteWebhook failed: Network request for 'deleteWebhook' failed!
[telegram] telegram setMyCommands failed: Network request for 'setMyCommands' failed!
[telegram] command sync failed: HttpError: Network request for 'setMyCommands' failed!
Environment & Context:

Model: NVIDIA Nemotron 3 Super 120B

OpenClaw version: 2026.3.11

Attempted fixes: > - Set NO_PROXY=api.telegram.org

Set NODE_TLS_REJECT_UNAUTHORIZED=0

Added ~/.local/bin to PATH to ensure openshell is visible.

Verified token is valid via curl (which returns 403 Forbidden through the default proxy, but works when the proxy is bypassed if the tunnel allows).

Question:
Is the Telegram bridge intentionally disabled in this sandbox tier, or is there a specific baseUrl or proxy configuration required for the undici fetcher to reach api.telegram.org

[zache-coder]

run the command
openshell term

and then you can see that 1 ore more rules are pending
you need to approve them, and then it works

[ericksoa]

The host-side Telegram bridge has been replaced by in-sandbox messaging via PR #1081. The network request failures described here were specific to the old bridge architecture. Closing as resolved by the architecture change.