writing-great-skills

Provides vocabulary and principles for writing predictable agent skills.

Updated Aug 2, 2026
One-click install
npx skills add https://github.com/mikalv/pi-extensions --skill writing-great-skills-mikalv
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: writing-great-skills
Source: https://github.com/mikalv/pi-extensions/tree/main/packages/pi-atelier/.agents/skills/writing-great-skills
Command: npx skills add https://github.com/mikalv/pi-extensions --skill writing-great-skills-mikalv

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Agent skills often behave inconsistently across runs because their instructions are bloated, duplicated, or poorly triggered. This Skill supplies a shared vocabulary and set of principles for writing and editing skills so the agent follows the same process every run. ## Core Features & Use Cases - Invocation Design: Explains the trade-off between model-invoked skills (description kept, agent-discoverable, context load) and user-invoked skills (description stripped, zero context load, human-triggered). - Information Hierarchy: Defines how to rank content into steps, in-file reference, and disclosed reference behind context pointers, with progressive disclosure and co-location guidance. - Failure Mode Diagnosis: Catalogs failure modes such as premature completion, duplication, sediment, sprawl, no-ops, and negation, each paired with its cure. - Use Case: When editing a skill whose agent keeps rushing steps, consult the failure modes section to sharpen completion criteria or split the sequence to hide post-completion steps. ## Quick Start Ask the agent to review your SKILL.md using the writing-great-skills principles and identify failure modes to fix.

Frequently Asked Questions about writing-great-skills

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write a good agent skill description?▼

Front-load the skill's leading word, list one trigger per branch, and cut identity already stated in the body. Synonyms that rename a single branch count as duplication and should be collapsed into one trigger phrase.

When should a skill be model-invoked vs user-invoked?▼

Choose model-invocation only when the agent or another skill must reach it autonomously, since the always-loaded description costs context load. If it only ever fires by hand, set disable-model-invocation and pay zero context load.

What is progressive disclosure in agent skills?▼

Progressive disclosure moves reference material out of SKILL.md into a linked file reached by a context pointer, keeping the top of the file legible. Inline what every branch needs and disclose what only some branches reach.

Why does my agent rush through skill steps?▼

Rushing is premature completion, caused by vague completion criteria and visible post-completion steps pulling attention forward. Sharpen the completion criterion first; only split the sequence to hide later steps if the criterion is irreducibly fuzzy.

What are leading words in skill writing?▼

Leading words are compact pretrained concepts like 'fog of war' or 'tracer bullets' that anchor agent behavior in few tokens by recruiting existing priors. They improve invocation reliability in descriptions and execution consistency in the body.

How do I keep a skill from bloating over time?▼

Check every line for relevance, hunt no-ops sentence by sentence, and keep each meaning in a single source of truth. Sediment accumulates because adding feels safe, so pruning must be an active discipline.