openclaw - 💡(How to fix) Fix [Feature]: Workspace file browser/editor in Control UI dashboard [1 participants]

Summary

It would be useful to browse and edit workspace files (USER.md, MEMORY.md, memory/*.md, etc.) directly from the WebChat dashboard, instead of needing to SSH in or ask the agent to make small edits. A simple file tree + text editor would save a lot of unnecessary AI turns for quick tweaks.

Problem to solve

OpenClaw agents rely heavily on workspace files (USER.md, MEMORY.md, SOUL.md, IDENTITY.md, TOOLS.md, HEARTBEAT.md, memory/*.md, etc.) for persistent memory, user context, and task configuration. Currently, the only ways to view or edit these files are:

1. SSH / terminal access — requires technical comfort and knowing the file paths
2. Asking the agent — wastes an entire AI turn (and tokens/cost) for simple changes like fixing a typo, adding a name, or tweaking a heartbeat task

For non-technical users or quick edits, neither option is efficient. There's no way to see what the agent "knows" about you without asking it, and no way to correct something without burning an API call.

Proposed solution

Add a workspace file browser and editor to the Control UI (WebChat dashboard):

• File tree panel — shows the workspace directory structure (~/.openclaw/workspace/) with expandable folders (e.g., memory/) • Built-in text editor — click any .md, .json, .csv, or .txt file to view and edit it inline, with a save button • Quick access shortcuts — pin commonly used files like USER.md, MEMORY.md, HEARTBEAT.md, and SOUL.md for one-click access • Read-only option — optionally allow viewing without editing for users who want to inspect but not accidentally modify

This would let users manage their agent's memory, preferences, and tasks directly — no terminal needed, no wasted AI turns. It keeps the human in control of what the agent knows, which aligns well with OpenClaw's transparency philosophy.

Alternatives considered

No response

Impact

• Reduced API costs — Simple edits like adding a contact name, fixing a typo, or adjusting a heartbeat task currently cost a full AI turn. A file editor eliminates these unnecessary token burns, which adds up fast at scale. • Better user experience for non-technical users — Not everyone is comfortable with SSH or terminal editors. A browser-based UI makes workspace files accessible to all users, lowering the barrier to entry. • Faster iteration — Users can tweak agent behavior (SOUL.md), memory (MEMORY.md), tool configs (TOOLS.md), or scheduled tasks (HEARTBEAT.md) in seconds instead of waiting for an agent round-trip. • Transparency and trust — Users can see exactly what the agent has stored about them at a glance, reinforcing OpenClaw's local-first, user-owned-data philosophy. • Debugging made easy — When the agent behaves unexpectedly, users can quickly inspect workspace files to understand why, rather than asking the agent to read files back to them (which itself costs tokens).

Evidence/examples

No response

Additional information

No response