writing-core

Routes drafting tasks to scenario writing skills and gates prose with a Vale-based checker.

1|Updated Mar 24, 2026
One-click install
npx skills add https://github.com/SgtPooki/skills --skill writing-core-sgtpooki
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: writing-core
Source: https://github.com/SgtPooki/skills/tree/main/skills/writing-core
Command: npx skills add https://github.com/SgtPooki/skills --skill writing-core-sgtpooki

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? Agent-drafted prose often carries AI slop, fabricated specifics, vague claims, and formulaic structure. This Skill provides a shared quality layer for any human-facing text — docs, PR descriptions, issues, specs, release notes, announcements — by routing each draft to the correct scenario writing skill and enforcing non-negotiable rules before delivery. ## Core Features & Use Cases - Scenario routing: Maps the artifact type to exactly one scenario skill (github-writing, writing-docs, writing-spec, writing-community, writing-marketing) with explicit edge-case rules. - Mandatory quality gate: Runs scripts/writingcheck.py on every draft, combining Vale lexical rules (Slop, Google, Marketing styles) with structural checks for single-bullet lists, closing summaries, thin sections, and unattributed statistics. - Composition pipeline: Sequences base drafting, optional persuasion pass, and author-voice overlay as distinct edits with defined precedence rules. - Use Case: When asked to write a README or PR description, the agent routes to the matching scenario skill, drafts with BLUF structure, runs the revision pass from references/, and gates the result through writingcheck.py before presenting it. ## Quick Start Ask the agent to write a PR description or README and have it route through writing-core, then run writingcheck.py on the draft before delivering it.

Frequently Asked Questions about writing-core

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

FAQPage Schema
How do I check AI-drafted text for slop words and filler?▼

Run scripts/writingcheck.py with a scenario name and the draft file or piped stdin text. It flags banned words like delve and seamless, hedging phrases, formulaic transitions, and structural issues such as single-bullet lists and closing summary sections.

How does writing-core choose which writing skill to use?▼

It routes by artifact type: GitHub artifacts to github-writing, persistent docs to writing-docs, decision records to writing-spec, announcements to writing-community, and persuasion copy to writing-marketing. Artifact purpose beats delivery mechanism when the two conflict.

Does writingcheck.py work without Vale installed?▼

Yes. When Vale is missing it reports VALE_MISSING and falls back to reduced checks that apply the Slop word rules and structural checks directly. The reduced mode is always reported explicitly and never silently skips.

What does a PASS from the writing checker actually mean?▼

A PASS means no known mechanical smells were found. It does not verify factual claims — the no-fabricated-specifics rule must be discharged by the writer running commands, reading sources, or linking authorities.

What are the limitations of the writing checker?▼

The rules and word lists are English-only, and the built-in overlay supports only author-voice, not a plugin system. Marketing rules are skipped in pass mode, so the honesty checklist must be walked manually after a persuasion pass.