Skip to content
SkillModel-led2,313 tokens in use

writing-for-agents

A reference for writing any document an agent reads, from skills to AGENTS.md and CLAUDE.md, built on context pointers, the information hierarchy and pruning.

328,791 stars on the repository MIT Licensemattpocock/skills · skills/productivity/writing-for-agents
claude plugin install mattpocock-skills@claude-plugins-official
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing writing-for-agents costs

28tokens idle
2,313tokens in use
2,866tokens, up to

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Up to adds the files its instructions link. Counts are approximate, made offline with cl100k_base.

Model dependence Model-led

Script-ledMixedModel-led

Mostly instructions, so how well it works depends on the model that follows them.

This describes the skill's structure, how much of it is scripts. It does not say how well it works.

Touches

Nothing detected: no shell, network, file writes, git or credentials in its files.

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Good
  • Frontmatter12/15
  • Structure25/25
  • Examples3/15
  • Output10/15
  • Size15/15
  • Resources7/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Excellent

No findings.

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Excellent
  • Development activity96
  • Maintenance100
  • Community56
  • Maturity32
  • Momentum40

writing-for-agents is a model-invoked reference skill for writing documents an agent consumes: skills, AGENTS.md and CLAUDE.md files, and any doc reached by a pointer. It treats them as one kind of document, because the same levers make each predictable.

It covers how to word a context pointer, how to weigh context load against cognitive load, where each piece of content sits on the information hierarchy, and how to prune. A companion file, SKILL-MECHANICS.md, covers frontmatter, invocation choice and router skills.

What You Get

  • A model-invoked skill that loads when you create or edit skills, AGENTS.md or CLAUDE.md
  • Rules for wording context pointers, with one trigger per branch
  • A framework for placing content in the file, in a sibling file or behind a pointer
  • A pruning checklist covering duplication, stale lines and no-op sentences
  • SKILL-MECHANICS.md for frontmatter, invocation choice and router skills

Common Use Cases

  • Writing a new skill and choosing its description wording
  • Trimming a bloated AGENTS.md or CLAUDE.md
  • Deciding whether to split a long skill by sequence or by invocation

What Makes It Unique

Most skill-authoring guides cover file format. This one covers the writing itself: how pointer wording decides when material loads, how completion criteria prevent premature endings, and how leading words replace restated passages. It also separates the context load on the agent from the cognitive load on the human.

Under The Hood

The skill is two Markdown files and no scripts. SKILL.md holds the reference, with a description that tells the agent to load it when creating or editing skills or instruction files. SKILL-MECHANICS.md sits beside it and loads when the document being written is a skill. The skill reads and edits nothing by itself and runs no commands.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers