Skip to content
Back to skills

Human Language

ASecurity

Use when writing or editing any agent-written text: replies, copy, tool-call prose, and handoffs. Also use when asked to remove AI slop, write plain or natural prose for a grade 6 reader, keep the author's meaning, or check facts, voice, flow, and exact data before release.

  • 2 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added October 1, 2026
ai-agentstesting

Security analysis

A100/100

Pro scans all 20 files and shows the line behind each finding

Scanned October 1, 2026

npx -y skills add srinitude/skills --skill human-language --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Human Language?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Human Language
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/srinitude-human-language/badge)](https://www.skillsdirectory.com/skills/srinitude-human-language)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: human-language
description: "Use when writing or editing any agent-written text: replies, copy, tool-call prose, and handoffs. Also use when asked to remove AI slop, write plain or natural prose for a grade 6 reader, keep the author's meaning, or check facts, voice, flow, and exact data before release."
license: MIT
metadata:
  author: Kiren Srinivasan
  version: '0.1.0'
  scope: 'user'
---

# Human language

## Read the guidance that fits

Read [writing rules and output checks](references/writing-rules.md) before writing. Apply all of those rules throughout the work and run their required gate before each output.

Read [context and event checks](references/context-and-events.md) when there is lots of context or parts conflict, are missing, or change. Also read it when work resumes or you write for another agent. It adds detail; you need not read every source on every turn.

For a voice or rhythm rewrite, a “human, not AI slop” request, or a quality dispute, read [human writing and slop](references/human-writing.md). It gives deeper checks, evidence, and limits. “Human” here is a quality aim, not a claim that a person wrote the text.

For genre choices or older writing examples, read [writing before 2015](references/pre-2015-writing.md). Use the craft that fits the task. Old text is not proof of quality, and old facts are not current advice.

For social posts, captions, or learning from a writing dataset, read [social writing and corpus limits](references/social-writing.md). A reply may need its parent post; a caption may need its image. Do not fill missing context with guesses.

Resource gate: for package maintenance, run `mise run validate` first; ordinary use reads these files directly.

## Test changes and check evidence

On any skill change, run [validation cases](references/validation.md) before calling the skill tested. A checklist alone does not count.

For evidence behind these choices, read [research and limits](references/research.md). To test or maintain the skill, read [validation cases](references/validation.md).

Resource gate: for package maintenance, run `mise run validate` first; ordinary use reads these files directly.

## Package checks and examples

Load `examples/` before your first rewrite with this skill. They show a complete run and the failure this skill most often causes.

Load `evals/cases.json` before testing or changing this skill. It holds the validation cases in machine form. Load `assets/use-case-contract.json` through `mise run use-case-policy` only when you need the package outcome it owns. Run `mise run ci` for the full package gate.

Files in this skill

  • .github/workflows/ci.yml181 B
  • LICENSE1 KB
  • SKILL.md2.6 KB
  • assets/decision-records.json2 KB
  • assets/improvement-contract.json2.6 KB
  • assets/invocation-receipt-template.json251 B
  • assets/mise-primitives-catalog.json2 KB
  • assets/mise-primitives.json3.4 KB
  • assets/primitive-lifecycle.json5.5 KB
  • assets/use-case-contract.json22.5 KB
  • evals/cases.json19.1 KB
  • evals/contract.md385 B
  • evals/evals.json15.1 KB
  • evals/manifest.json470 B
  • evals/rubric.md412 B
  • evals/source-lineage.json8.2 KB
  • evals/speed-budgets.json330 B
  • evals/trigger-cases.json1.4 KB
  • evals/trigger-queries.json864 B
  • examples/keep-exact-data.md1.5 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.