intrusive-thoughts
by kittleik
Web & Frontend DevelopmentBrief
Autonomous AI consciousness starter kit.
Description
What This Skill Does
A framework that adds autonomous behavior to AI agents: mood-driven scheduling, multi-store memory with Ebbinghaus decay, trust learning, self-evolution, and a local web dashboard. Agents check weather and news each morning to pick a mood, then schedule spontaneous activities throughout the day and overnight.
The entire system runs offline with no API keys required for core functionality, using only public RSS feeds and a free weather endpoint.
When to Use It
- Giving an AI agent a daily mood based on local weather and headlines
- Running overnight deep-work sessions for an agent while you sleep
- Reviewing agent mood history and activity stats on a local dashboard
- Configuring an agent's personality from archetype presets like Tinkerer or Night Owl
- Letting an agent learn when to act autonomously versus when to ask for permission
Example Workflow
User asks: Run today's morning mood ritual for my agent
- 1
Run set_mood.sh to fetch current weather from wttr.in and headlines from BBC and HackerNews RSS
- 2
Read moods.json to evaluate weather and news influence maps against available moods
- 3
Select a mood and write the result to today_mood.json
- 4
Run schedule_day.py to generate one-shot pop-in cron jobs for the day via the OpenClaw API
- 5
Send the user a message summarizing the chosen mood and today's schedule
Agent is in Curious mode with three pop-in activities queued for the afternoon
Requirements
- OPENAI_API_KEY environment variable (optional, for enhanced AI features)
- Telegram bot token in config.json (optional, for outbound notifications)
Installation
clawhub install kittleik/intrusive-thoughtsRun the CLI command, or paste the prompt into Claude Code, Cursor, Codex or whichever coding agent you use.