AGENT SKILL

time-audit

暂无简介

License 许可
MIT
最近更新
2026-09-04
Stars
★ 0
Forks
0

02 资源文件 3

03 SKILL.md 正文

TimeAudit — OpenClaw Skill Documentation

ADHD-friendly voice time tracker for the Rabbit R1. Log context switches in under 2 seconds. Zero-friction, voice-first, with smart categorization, distraction detection, and billable time export.


Overview

TimeAudit runs entirely on-device. OpenClaw is optional but unlocks two features:

Feature Without OpenClaw With OpenClaw
Voice log + local LLM parse Full support Full support
Distraction nudge + Pomodoro Full support Full support
Daily summary (spoken) Full support Full support
CSV timesheet export Queued (no-op) Exports to ~/timesheets/
Weekly reflection Falls back to local analysis Self-Hosted Backend AI insights

Setup

1. Install the Creation

Drop time-audit/ into your Rabbit R1 Creations directory and open it on the device.

2. OpenClaw (optional)

To enable CSV export and AI weekly reflections, run the OpenClaw skill bridge on your local machine (Self-Hosted Backend):

cd ~/openclaw
node index.js --skill time-audit --handler ./handlers/time-audit.js

3. self-hosted-gateway (optional)

Weekly reflections are generated by the self-hosted-gateway OpenClaw skill. Ensure self-hosted-gateway is registered and has an LLM API key configured.


Usage — PTT Controls

Action Result
Quick PTT press Start listening — say what you're switching to
Release PTT Stop listening, parse and log the switch
Long-press PTT (track screen) Speak daily summary
Long-press PTT (summary screen) Close summary
Scroll wheel up/down Scroll recent switches list
Long-press any switch row Open billing tag screen

Voice Commands

TimeAudit understands natural language. The R1 LLM parses intent:

"email"                              → Communication / Email
"deep work on the API refactor"      → Deep Work / API Refactor
"thinking about the Henderson proposal" → Planning / Henderson Proposal
"bathroom break"                     → Break / Break (not billed)
"lunch"                              → Break / Lunch
"standup meeting"                    → Meetings / Standup Meeting
"start a focus block"                → Pomodoro 25-minute timer
"show weekly reflection"             → self-hosted gateway weekly insights

Categories

Auto-assigned by LLM with local keyword fallback:

Category LED Color Keywords (sample)
Deep Work Green code, write, design, build, implement, research
Communication Blue email, slack, chat, reply, inbox
Meetings Amber meeting, call, standup, zoom, interview
Admin Purple organize, schedule, expense, invoice, file
Creative Pink brainstorm, sketch, prototype, art
Learning Teal course, tutorial, podcast, docs, study
Planning Amber plan, roadmap, strategy, goals, reflect
Break Teal lunch, coffee, bathroom, rest, gym
Unknown Off anything unmatched

Distraction Detector

If you log more than 4 context switches in any 90-minute window, TimeAudit:

  1. Shows the nudge screen with switch count
  2. Blinks LED red (4 pulses at 300ms)
  3. Speaks: "You've switched tasks N times in the last 90 minutes. Want to try a 25-minute focus block?"
  4. Offers two options: Focus 25m or Dismiss
  5. Dismiss has a 1-hour cooldown before nudging again

Pomodoro Integration

Trigger with voice: "focus block", "pomodoro", "start a focus block", or tap Focus 25m on the nudge screen.

  • 25-minute countdown with animated ring
  • LED: green → amber (10 min left) → red (5 min left)
  • Context switch logging is blocked during the block
  • On completion: speaks encouragement, auto-logs a Break, sets 30-min nudge cooldown
  • Cancel anytime: quick PTT press on the pomo screen

Screen Layout (track screen)

┌──────────────────────────────┐  ← 5px LED border
│ [Category badge]   [0:00]    │  ← current category + elapsed
│                              │
│ Current Task Name            │  ← large, 19px bold
│ (2-line max)                 │
│                              │
│ [Pie chart] Deep Work  34%  │  ← today's breakdown
│             Email      26%  │
│             Meetings   15%  │
│             Breaks     11%  │
│                              │
│ Recent Switches ─────────── │
│ ● Current Task      [running]│
│ ● Previous          [45m]    │
│ ● Earlier           [1h20m]  │
│ ● Even earlier      [30m]    │
└──────────────────────────────┘

OpenClaw Handler — Full Implementation

Save as ~/openclaw/handlers/time-audit.js:

/**
 * time-audit OpenClaw handler
 * Commands: export-csv
 */

const fs   = require('fs');
const path = require('path');
const os   = require('os');

const TIMESHEET_DIR = path.join(os.homedir(), 'timesheets');

module.exports = {
  skill: 'time-audit',

  async handle(command, params, storage) {
    switch (command) {

      case 'export-csv': {
        const { entries } = params;
        if (!entries?.length) {
          return { status: 'done', data: { error: 'No entries to export' } };
        }

        fs.mkdirSync(TIMESHEET_DIR, { recursive: true });

        const dateStr  = new Date().toISOString().slice(0, 10);
        const filename = `timesheet-${dateStr}.csv`;
        const filepath = path.join(TIMESHEET_DIR, filename);

        const header = 'Date,Task,Category,Start,End,Duration (min),Billable\n';
        const rows   = entries.map(e => [
          e.date,
          `"${e.task.replace(/"/g, '""')}"`,
          e.category,
          e.start,
          e.end,
          e.durationMin,
          e.billable ? 'Yes' : 'No',
        ].join(',')).join('\n');

        fs.writeFileSync(filepath, header + rows, 'utf8');

        // Also append to a monthly roll-up
        const monthFile = path.join(TIMESHEET_DIR, `timesheet-${dateStr.slice(0,7)}.csv`);
        const monthExists = fs.existsSync(monthFile);
        fs.appendFileSync(monthFile, (!monthExists ? header : '') + rows + '\n', 'utf8');

        return {
          status: 'done',
          data: {
            file: filepath,
            entriesExported: entries.length,
            totalMinutes: entries.reduce((s, e) => s + e.durationMin, 0),
          },
        };
      }

      default:
        return { status: 'done', data: { error: `Unknown command: ${command}` } };
    }
  },
};

OpenClaw registration

Add to your openclaw/config.json:

{
  "skills": [
    {
      "name": "time-audit",
      "handler": "./handlers/time-audit.js",
      "permissions": ["filesystem:write:~/timesheets/"]
    },
    {
      "name": "self-hosted-gateway",
      "handler": "./handlers/self-hosted-gateway.js",
      "permissions": ["llm:generate"]
    }
  ]
}

Storage Schema

All data lives in R1 Creation storage (on-device, no cloud).

ta:day:YYYY-M-D

{
  "entries": [
    {
      "task": "Email triage",
      "category": "Communication",
      "startMs": 1718000000000,
      "endMs":   1718001800000,
      "durationMs": 1800000,
      "billable": false,
      "hourBlock": 9
    }
  ],
  "savedAt": 1718010000000
}

ta:current

{
  "task": "API refactor",
  "category": "Deep Work",
  "startMs": 1718009000000,
  "billable": true
}

Persisted every 30 seconds and on beforeunload so the app survives restarts mid-task.


Accessibility

All state changes include:

  • LED color change — visible at a glance
  • TTS announcement — spoken confirmation ("Got it. Starting: Email triage")
  • Screen update — visual reinforcement

Core flow (PTT → speak → confirm) requires zero screen touch. Billable tagging requires a long-press on switch rows (touch only).

Input debounce: 300ms minimum on PTT to protect users with tremor.


Weekly Reflection — Self-Hosted Backend Prompt

The app sends this prompt to the self-hosted-gateway skill:

Weekly time audit: Deep Work: 8.2h, Communication: 4.1h, Meetings: 3.5h, Break: 2.3h.
Total: 18.1h.
Provide 2-3 concrete, actionable time management suggestions based on these patterns.
Be specific and encouraging. Max 80 words.

Example response:

Your deep work ratio (45%) is excellent. To protect it further, try batching all email and Slack into two 30-minute windows — morning and afternoon. Meetings are consuming 3.5h; see if any can become async. Your best day appears to be Tuesday — schedule your hardest deep work there.


Troubleshooting

Voice not triggering?

  • Ensure the R1 PTT button fires sideClick events in your creation context.
  • CreationVoiceHandler.postMessage('start') requires mic permission in the Rabbit OS.

LLM parse returning wrong category?

  • The local keyword fallback runs if the LLM call fails or returns malformed JSON.
  • Categories are re-assignable in the billing screen (future feature).

Distraction nudge not appearing?

  • Requires > 4 switches in 90 minutes AND state.screen === 'track' at the 2.5s delay.
  • After dismissal, there is a 1-hour cooldown.

CSV export failing?

  • OpenClaw must be running on your self-hosted gateway (node openclaw/index.js).
  • Check the ~/timesheets/ directory exists and is writable.
  • Verify time-audit skill is registered in openclaw/config.json.

04 同仓库更多 Skills

ajohnsonnow
ajohnsonnow/r1-warren

whisper-quiz

WhisperQuiz — voice-answered trivia game for Rabbit R1. Generates questions via the R1's built-in LLM, captures spoken answers via STT, and evaluates correctness with flexible LLM-based matching.

★ 0
ajohnsonnow
ajohnsonnow/r1-warren

walk-cast

Walking podcast production skill. Receives raw audio metadata and transcript from the R1, removes long pauses, applies noise reduction and normalization, generates AI chapter titles and episode metadata via LLM, and uploads the final episode to your configured podcast RSS host (Spotify for Podcasters / Anchor, Podbean, Buzzsprout, RSS.com, or any FTP-accessible feed).

★ 0
ajohnsonnow
ajohnsonnow/r1-warren

voice-novel

Local machine skill for VoiceNovel — handles story export to disk.

★ 0
ajohnsonnow
ajohnsonnow/r1-warren

voice-guide

Environment description and navigation assistant for blind and low-vision users. Analyzes camera frames via a vision LLM, reads visible text, identifies scene categories (people/vehicles/food/hazards/exits), performs 360° scans, retrieves GPS position, and sends emergency SMS alerts — all via audio feedback with zero screen dependency.

★ 0
ajohnsonnow
ajohnsonnow/r1-warren

voice-draft

Voice-driven document drafting for the Rabbit R1. Dictate stream-of-consciousness speech; the R1 formats it into Email, Blog Post, Report, Proposal, Meeting Notes, Creative Writing, Code Comments, Cover Letter, or Tweet Thread — then saves the result to a configured folder on your local machine via OpenClaw.

★ 0
ajohnsonnow
ajohnsonnow/r1-warren

voice-commit

Hands-free git and CI workflow voice controller for the Rabbit R1. Speak git/CI commands and hear results — commit, push, status, PR management, test runs, and deploy, all voice-driven via OpenClaw.

★ 0