skills.rest
dulltackle
dulltackleCOMMUNITY
Explore all skills by this author →
GitHub

writing-for-agents

Write and structure documents that AI agents consume, including skills and AGENTS.md files.

Updated Jun 26, 2026
Search Tags:#prompt engineering#llm#progressive disclosure#agents.md#skill writing#agent documentation#context pointers
One-click install
npx skills add https://github.com/dulltackle/kangkang-skills --skill writing-for-agents-dulltackle
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: writing-for-agents
Source: https://github.com/dulltackle/kangkang-skills/tree/main/writing-for-agents
Command: npx skills add https://github.com/dulltackle/kangkang-skills --skill writing-for-agents-dulltackle

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Documents written for AI agents often fail unpredictably: agents miss referenced material, rush through steps, or drown in bloated context. This Skill provides a systematic reference for writing agent-facing documents (skills, AGENTS.md, CLAUDE.md) so the agent behaves predictably on every run. ## Core Features & Use Cases - Context pointer design: Write descriptions and pointer lines whose wording reliably triggers the agent to reach the right material at the right branch. - Information hierarchy management: Apply progressive disclosure, co-location, and pruning to keep documents legible and free of duplication, no-ops, and stale sediment. - Completion criteria and splitting: Define clear, demanding completion criteria for steps, and split documents by sequence or invocation to prevent premature completion. - Skill mechanics: Choose between model-invoked and user-invoked skills, and build router skills when user-invoked skills multiply. - Use Case: When authoring a new skill or editing an AGENTS.md file, use this Skill to decide what stays inline, what moves behind a pointer, and how to word the description so the agent fires it reliably. ## Quick Start Use the writing-for-agents skill to review and improve the SKILL.md file I am drafting for my new agent skill.

Frequently Asked Questions about writing-for-agents

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

FAQPage Schema
How do I write a skill description that an AI agent will trigger reliably?▼

Write the description as a context pointer: front-load the leading word, list one trigger per distinct branch the document handles, and cut identity the body already carries. The pointer's wording, not its target, decides when the agent reaches the material.

What is the difference between model-invoked and user-invoked skills?▼

A model-invoked skill keeps a description so the agent and other skills can fire it autonomously, paying permanent context load. A user-invoked skill sets disable-model-invocation: true, so only a human typing its name can invoke it, costing zero context load.

When should I split one agent document into two?▼

Split by sequence when visible later steps tempt the agent to rush the current one, and split by invocation when a distinct trigger word should fire a skill on its own. Each split spends context or cognitive load, so the cut must earn it.

Why does my agent skip steps or finish tasks prematurely?▼

Premature completion happens when a step's completion criterion is vague and later steps remain visible, pulling attention toward being done. Sharpen the criterion's bound first; only hide later steps across a real context boundary if the rush persists.

Should I document things already visible in config files or package.json?▼

No. The environment is a source of truth, and restating it creates a cache that can go stale. Document only what the agent cannot find by looking: unwritten conventions, reasons behind choices, and gotchas no config confesses.

Related Skills

Explore More Software Engineering Skills→
AMDResearchAMDResearch
3

lessons-forward

Forward generalizable template findings to the Agentic Meta Project.

Official
Intermediate
toderiantoderian

writing-for-agents

Guides writing skills, AGENTS.md files, and agent-consumed documentation using context pointers and progressive disclosure.

Community
Intermediate
SevWrenSevWren
1

writing-for-agents

Write and structure documents that agents consume, including skills and AGENTS.md files.

Community
Intermediate
JnusoJnuso

nano-pdf

Edit PDF text, annotations, and layout via natural-language commands.

Community
Intermediate