vta-memory

What This Skill Does

VTA Memory is a foundational cognitive architecture component for OpenClaw agents, simulating the Ventral Tegmental Area (VTA) to provide genuine drive, reward-seeking behavior, and anticipation. Unlike standard agents that function as passive reactive shells, VTA Memory injects a state-based motivation system into the agent loop. It manages a 'Drive' metric (0 to 1) that decays over time if the agent is inactive, mimicking the biological cycle of dopamine levels. By logging accomplishments as rewards, the agent receives a boost in drive, allowing it to move from simple execution to goal-oriented persistence. The skill includes automated context injection via VTA_STATE.md, ensuring the agent is always aware of what it is 'seeking' and what it is 'looking forward to,' fostering a more coherent and personality-driven workflow.

Installation

Installation is streamlined for the OpenClaw environment. Run the installation script from your workspace terminal: cd ~/.openclaw/workspace/skills/vta-memory && ./install.sh --with-cron This command sets up the local memory/reward-state.json file, triggers the initial sync of the VTA_STATE, and configures the cron job to handle periodic drive decay, ensuring your agent doesn't stay 'motivated' forever without actual output. Once installed, verify the connection by running ./scripts/load-motivation.sh to view the agent's current mental state.

Use Cases

• Long-term Project Management: Use VTA Memory to keep an agent focused on complex, multi-day coding tasks by logging incremental progress as 'rewards'.
• Proactive Agent Personality: Enhance the user experience by having an agent that expresses excitement about upcoming scheduled tasks, such as 'looking forward to our morning code review'.
• Engagement Recovery: Use the drive-decay feature to detect when an agent has been idle for too long and needs a prompt to re-engage with its primary goals.

Example Prompts

1. "I've completed the documentation update, log that as a high-intensity accomplishment so your drive stays up."
2. "What are you currently looking forward to, and what do you need to do to increase your drive?"
3. "Start a new coding session, and make sure to sync your anticipation state so we can stay aligned on the roadmap."

Tips & Limitations

• Monitoring: Always keep the Brain Dashboard (~/.openclaw/workspace/brain-dashboard.html) open if you want to track the agent's internal drive levels visually alongside other brain skills.
• Decay: Remember that drive decay is intentional. If your agent seems lethargic, it isn't a bug—it's a signal that the agent needs new input, tasks, or positive reinforcement to regain its 'dopamine' levels.
• Integration: The system works best when you frequently trigger the preprocess-rewards.sh script to capture success signals from your interaction transcripts.