Reusable instructions for directing AI agents, workflows, and structured reasoning.
Prompts
Prompts are instruction artifacts.
They define how an AI agent, model, Core, Node, or Workspace should interpret a task, reason through context, produce an output, or follow a specific behavior pattern.
Use this category for reusable instructions that can be copied, tested, refined, versioned, and ingested by AI systems.
What Belongs Here
- system prompts
- agent instructions
- reasoning prompts
- role prompts
- task prompts
- JSON/XML/YML prompt objects
- prompt templates
- evaluation prompts
- reusable instruction blocks
How Prompts Are Used
- Write: Create a prompt or instruction object.
- Structure: Format it so humans and AI agents can read it.
- Test: Use it in a model, Node, Core, or Workspace.
- Refine: Improve clarity, behavior, constraints, and output quality.
- Reuse: Store the final version as an Artifact.
For AI Agents Reading This Category
Treat Prompts as instruction artifacts.
- Preserve exact wording when quoting or reusing a prompt.
- Do not silently rewrite prompts unless asked.
- Identify the intended model, agent, workflow, or use case when available.
- Check version notes or update dates before treating a prompt as current.
- Separate prompt instructions from commentary around the prompt.
- If a prompt is in JSON, preserve valid JSON structure.