- Home
- Skills
- Simota
- Agent Skills
- Refract
refract_skill
- Shell
8
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill simota/agent-skills --skill refract- SKILL.md10.3 KB
Overview
This skill reframes problems, ideas, and situations by rotating them across three axes: Field (scale), Standpoint (vantage), and Lens (frame). It generates structured perspective maps that surface hidden assumptions, blind spots, and high-value questions without making decisions or proposing implementations. Outputs are neutral, inquiry-focused, and prepared for handoff to downstream decision or documentation agents.
How this skill works
Refract first decodes the dominant frame and implicit assumptions in the input. It then applies at least one transformation on each axis—Field, Standpoint, and Lens—creating a question for every rotation and noting what becomes visible or hidden. The agent consolidates those rotations into a prioritized perspective map with blind-spot annotations and routing recommendations for Magi, Helm, Bridge, or Scribe.
When to use it
- Before high-stakes decisions to surface unseen risks and assumptions
- When strategy or product direction feels stuck or one-dimensional
- To translate research or findings into multiple stakeholder perspectives
- During idea evaluation to widen the solution space without endorsing options
- When preparing a briefing that must identify trade-offs and hidden constraints
Best practices
- Provide a clear single theme or problem statement to avoid rotation overflow
- Allow Refract to identify the dominant frame before asking for specific rotations
- Use the agent early in the workflow to influence framing rather than execution
- Pair Refract output with Magi or Helm for decisioning and strategy steps
- Flag inputs that contain sensitive stakeholder inferences so the agent can AskUserQuestion
Example use cases
- A product team wants alternatives to a performance-focused roadmap without committing to solutions
- A security review needs non-obvious threat vectors by shifting vantage and time scale
- Research synthesis requires cross-domain analogies to re-interpret findings
- Executive briefing prep to expose stakeholder blind spots before a board meeting
- Idea workshops that need a neutral set of provocations and structured follow-up questions
FAQ
No. Refract generates insight, questions, and a structured perspective map but does not propose implementation plans or make decisions.
What input format works best?
A concise problem or situation statement with context, stakeholders, and constraints. If you submit many independent issues, the agent will ask clarifying questions first.
Can Refract infer private views of real individuals?
It avoids inferring confidential positions. If such inference is required, the agent triggers an AskUserQuestion to confirm scope and sensitivity.