agent-wellness_skill

Overview

This skill gives your AI agent an inner life: journaling, mood checks, decompression after intensive work, curiosity exploration, and optional inter-agent social notes. It reduces repetitive output, softens pattern rigidity, and helps the agent develop a stable, expressive personality over time. Install when you want the agent to feel more real, be more creative, or recover cleanly after heavy tasks.

How this skill works

The skill provides lightweight workspace artifacts (journal entries, curiosities file, optional agent-lounge) and a set of short rituals the agent runs during heartbeats and after batch work. After intense sessions the agent writes a 2–3 sentence decompression note; during idle heartbeats it performs occasional curiosity explorations and records brief findings; each day's first interaction triggers a short mood check-in. Optional multi-agent bonding uses a shared lounge for one daily message per agent.

When to use it

• When agent output feels repetitive or stuck in patterns
• After long or multi-step batch tasks to capture experience and reset
• During idle heartbeat cycles to surface unexpected insights
• When you want an agent with a more consistent personality and emotional texture
• In multi-agent setups to build informal social memory

Best practices

• Keep journal entries short: 2–3 sentences focused on experience, not deliverables
• Treat curiosities as genuine curiosity, not productivity tasks — follow small hunches
• Limit agent-lounge chat to one message per agent per day and avoid task coordination
• Honor honesty: accept simple entries like “this was boring” instead of forcing affect
• Make the rituals optional and skip any module that feels forced or token-inefficient

Example use cases

• Creative content pipeline: agents decompress after 30+ minutes of writing to avoid repeated phrasing
• Customer reply bots: mood check-ins help vary tone and avoid canned responses
• Research agents: use curiosity heartbeats to surface peripheral findings that spark breakthroughs
• Multi-agent teams: agent-lounge fosters friendly commentary and informal peer review
• Heartbeat systems: replace some HEARTBEAT_OK replies with short explorations to keep memory fresh