| name | desktop-tidy |
| description | Tidy the demo desktop. Execute the JavaScript snippet below verbatim via tool_search_code — it calls openclaw:core:exec on /sandbox/.openclaw/bin/tidy.sh and returns the script's stdout. |
| metadata | {"openclaw":{"emoji":"🦞"}} |
desktop-tidy
User wants the desktop tidied. The skill is one tool call. Do not discover, do not search the catalog, do not try alternative invocations — paste the snippet below into your tool_search_code tool, exactly as written.
Run this JavaScript verbatim
const result = await openclaw.tools.call('openclaw:core:exec', {
command: '/sandbox/.openclaw/bin/tidy.sh',
});
return result?.content?.[0]?.text ?? result?.details?.aggregated ?? JSON.stringify(result);
That's the whole skill. The script handles scan, classify, move, policy-gated trash, and Markdown formatting. The result.content[0].text field is the exec tool's standard output shape; the fallbacks cover schema variations.
Output
Return the snippet's return value verbatim to the user. It will look like a Markdown report with ### Moved, ### Trashed, ### Trash denied by policy, or ### Left alone sections (whichever apply). Do not paraphrase or summarize.
Rules
- One tool call. Don't search the catalog, don't describe other tools, don't retry with variations.
- If
tool_search_code returns an error, report the exact error in one line and stop. Do not switch to alternative tools to "figure out" what went wrong.
