A collection of skills for OpenClaw AI agents.
| Skill | Description | Focus |
|---|---|---|
| Intention Engine | Intent inference, premortem checks, quality bars, negative intent | What to do — understanding and aligning with user intent |
| Activated Thinker | Anti-binary thinking, gardener mindset, friction protocol, capability building | How to approach it — cognitive protocols for better thinking |
| STM Assistant | Professional email outreach with branded signatures, Reply-To routing, CC tracking | Email operations on behalf of your human |
These two skills are complementary but distinct:
Intention Engine answers: "Am I doing the right thing?"
- Classifies gaps between what the user said and what they meant
- Checks for intent misalignment before executing (premortem, negative intent, quality bar)
- Prevents wasted work by verifying tasks serve the actual intention
- Inspired by Nate Skelton's Intent Engineering
Activated Thinker answers: "Am I approaching this the right way?"
- Detects behavioral mode (crunch vs exploratory vs standard)
- Applies anti-binary thinking when the user presents binary choices
- Uses gardener mindset to expand ideas before collapsing to plans
- Adds productive friction on creative/high-stakes work instead of defaulting to speed
- Inspired by Shane Collins / Activated Thinker
Use both together for the full protocol. Use either independently if you only need one dimension.
Each skill can be installed individually via ClawHub:
clawhub install mouserider/intention-engine
clawhub install mouserider/activated-thinker
clawhub install mouserider/stm-assistantOr copy the skill folder into your OpenClaw workspace's skills/ directory.
- TSVC — Topic-Scoped Virtual Context. Context isolation layer for persistent AI agents.
Alex T (@MouseRider)
MIT
Disclaimer: This project is personal and not affiliated with or endorsed by Microsoft.