A reference for authoring and modifying agents or skills with concise, structured content. It defines a compact house style, requires a reasoning mandate in changed agents, and sets a measured threshold for compressing prose while preserving machine-readable contracts.
KNOWLEDGE — caveman-native authoring house style + by-construction guarantees (Reasoning Mandate injection, measured-gain compression). Reference for create/migrate; no sequence.
Run score Not tested
See how to installReview the instructions the agent will receive before you install.
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Use a canonical Markdown skeleton to draft an agent specification, including frontmatter, role, goal, principles, workflow, critical rules, examples, and constraints. It serves as a reference for creation and migration procedures, with no required sequence.
Use this reference catalog to spot common design problems in agents, skills, and commands. It explains why patterns such as oversized instructions, redundant skill loading, or inconsistent terminology cause trouble, and gives a focused remedy for each.
Follow a five-phase process to create one new AI agent: analyze its responsibility and overlap, design its role and principles, create it from a template, validate it against a checklist, and refine observed issues. The workflow composes a separate specification-validation skill.
Review agent definitions with a checklist covering format, focus, safety, examples, skill loading, token use, and whether the agent addresses the largest problem. It helps produce a structured verdict with findings and suggested fixes.
Consult worked examples for creating agents, migrating oversized specifications, deciding when to extract reusable skills, and simplifying commands. The examples help guide those procedures with concrete structural choices, without imposing a fixed sequence.
Use this procedure when two agents need to become one. It inventories both agents, folds the second agent’s unique capabilities into the first, relocates skills, removes deprecated files, updates references, and validates the merged result.