openclaw - 💡(How to fix) Fix [Design Feedback] Exec approval UX breaks Telegram mobile-first experience [3 comments, 4 participants]

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…
GitHub stats
openclaw/openclaw#59102Fetched 2026-04-08 02:28:35
View on GitHub
Comments
3
Participants
4
Timeline
4
Reactions
0
Timeline (top)
commented ×3subscribed ×1
RAW_BUFFERClick to expand / collapse

Problem 1: Approval and output are decoupled

When approvals.exec.mode: "targets" is set, exec results are sent directly to the target channel (Telegram), bypassing the agent. The user sees raw terminal output in Telegram with no formatting, while the approval UI lives in Control UI on a different surface.

Problem 2: Mobile approval is painful

The fallback approval UI (Control UI) is not mobile-friendly. Long commands cannot be fully scrolled, making it impossible to reach confirm/deny buttons.

Problem 3: Config is fragmented and opaque

Approval behavior is controlled by three separate mechanisms: approvals.exec in openclaw.json, tools.exec, and ~/.openclaw/exec-approvals.json.

Suggested fixes

  1. Route exec output back through the agent for formatting before sending to the user
  2. Add a session-scoped approval mode so users approve once per conversation rather than per command
  3. Make approval UI fully usable on mobile
  4. Consolidate exec approval configuration into one clear place
  5. Default to session-scoped approval for Telegram/DM use cases

The core value of a personal AI agent is being reachable from mobile. A chat-first assistant that requires a desktop browser to approve commands defeats its own purpose.

extent analysis

TL;DR

Route exec output back through the agent for formatting before sending to the user to address the decoupling of approval and output.

Guidance

  • To address Problem 1, consider modifying the approvals.exec.mode configuration to route output through the agent for formatting, ensuring users receive formatted output in Telegram.
  • For Problem 2, redesign the approval UI in Control UI to be mobile-friendly, allowing for full scrolling of long commands and easy access to confirm/deny buttons.
  • To simplify configuration (Problem 3), consolidate approval behavior settings into a single location, such as openclaw.json, to reduce fragmentation and improve transparency.
  • Consider implementing a session-scoped approval mode to enhance user experience, especially on mobile devices.

Example

No specific code example is provided due to the lack of detailed implementation details in the issue description.

Notes

The suggested fixes aim to improve the user experience, especially on mobile devices, and to simplify the configuration of approval behaviors. However, the exact implementation details may vary depending on the specific requirements and constraints of the system.

Recommendation

Apply workaround: Implement a session-scoped approval mode and route exec output through the agent for formatting to improve the mobile user experience and simplify approval configuration. This approach addresses multiple problems outlined in the issue and aligns with the core value of a personal AI agent being reachable and usable from mobile devices.

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…

Still need to ship something?

×6

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

Back to top recommendations

TRENDING