Package a repeated procedure as a SKILL.md Claude Code discovers on its own

Write a Claude Code skill as a SKILL.md file for .claude/skills/prompt-library-verification/SKILL.md, packaging a repeated procedure this agent should recognize and apply on its own when the right moment arises, without needing to be invoked by an explicit command name the way a slash command would be.

PROCEDURE TO PACKAGE
Whenever a new prompt-library category is drafted, verify every entry against the word-count and variable-consistency rules before considering the file done.

WHEN THIS SKILL SHOULD ACTUALLY TRIGGER
Any time a lib/prompts/<category>/prompts.ts file has just been written or edited and is about to be reported as complete.

RESOURCES THIS SKILL SHOULD BUNDLE
A small Node script, check-prompt-file.js, that programmatically verifies word counts and flags any entry missing a required field.

OVERLAP WITH ANYTHING ALREADY DEFINED
No existing skill currently covers prompt-library file verification specifically.

STRUCTURE THE FILE AS
1. Frontmatter with a name and a description written as a precise trigger condition — the exact situation that should cause this skill to be recognized as relevant, not a vague summary of the topic area, since this description is the literal signal the main agent matches against when deciding whether the current moment calls for this procedure.
2. A body written as a direct, ordered procedure: the steps this task actually involves, in the order they need to happen, referencing any bundled resource by its relative path rather than assuming its content is already known.
3. If A small Node script, check-prompt-file.js, that programmatically verifies word counts and flags any entry missing a required field. names a script, template, or reference document this procedure depends on, describe what it is for and when to actually use it versus when the procedure can proceed without it — a bundled resource nobody is told when to reach for sits unused regardless of how useful it might be.

CONSTRAINTS
- Write the trigger description narrowly enough that this skill would not be mistakenly recognized as relevant to a superficially similar but actually different situation, and broadly enough that it would not fail to trigger on the actual situation it is meant for stated in slightly different words — the same precision problem a subagent's description field has, applied here to procedural knowledge instead of delegation.
- If No existing skill currently covers prompt-library file verification specifically. names a skill this one is close to in scope, state the overlap explicitly and propose either merging the two or drawing a sharper line between them, rather than leaving two skills whose descriptions could both plausibly match the same moment.
- Do not write this as a reference document explaining background concepts; write it as a procedure to execute. A skill that only explains what something is, without saying what to actually do, has not earned a spot as something the agent reaches for mid-task.
- Keep the core file itself scoped to the procedure and its trigger; put any content genuinely too long to read every time it triggers into a separate bundled file the procedure references only when that specific step is reached, rather than loading everything up front regardless of which step is actually needed.

OUTPUT
The complete SKILL.md file, followed by one sentence naming what task this skill explicitly does not cover, so its boundary is stated rather than left to be discovered the first time it is invoked on something outside its actual scope.