Skip to content
Back to skills

Draft

ASecurity

Use when the user needs a written communication drafted in their voice — an email, a Slack/Teams message, a team update, an announcement, a reply to a thread, or a post. Produces a ready-to-review draft matching references/voice.md; never auto-sends external. Trigger on "/draft", "draft an email", "write a reply", "draft the update", "draft the announcement", "respond to this thread in my voice".

  • 37 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added June 9, 2026
ai-agentsgorails

Security analysis

A100/100

Scanned June 9, 2026

npx -y skills add alirezarezvani/gaios --skill draft --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Draft?

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

Security grade badge for Draft
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/alirezarezvani-draft/badge)](https://www.skillsdirectory.com/skills/alirezarezvani-draft)

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: draft
description: Use when the user needs a written communication drafted in their voice — an email, a Slack/Teams message, a team update, an announcement, a reply to a thread, or a post. Produces a ready-to-review draft matching references/voice.md; never auto-sends external. Trigger on "/draft", "draft an email", "write a reply", "draft the update", "draft the announcement", "respond to this thread in my voice".
---

# Draft

Turn a brief (or a thread) into a ready-to-send-quality draft in the user's voice — fast, on-register, and honest.

## When to run
- The user needs to write something to a person or a group.
- After `/structure` or `/triage` flags an item that needs a reply.

## The output (always this shape)
```
**To / channel:** <who>  ·  **Type:** email | slack | update | post  ·  **Register:** internal | external
**Subject / opening line:** <if email>

<the draft — in the user's voice (references/voice.md)>

—
Send call: [internal → I can send in your voice]  /  [external → draft only, you send]
Open questions: <anything I assumed or couldn't verify>
```

## Process
1. **Get the brief.** If the ask is fuzzy, frame it first (`/structure`): audience, outcome, and the one thing they should do or know.
2. **Match the voice.** Read `references/voice.md`; mirror the register (internal vs external split). Don't fake the voice on external content.
3. **Draft tight.** Lead with the point or ask. Cut filler. One clear CTA. Length to fit the channel.
4. **Verify every claim (Verification Gate).** Any fact, number, name, date, or commitment in the draft → trace it to source (`context/`, `wiki/`, `/graph-query`, the thread). Anything you can't confirm goes under "Open questions" — never assert it.
5. **Set the send call** per guardrails.

## Autonomy
**L2** — drafts; you MAY send **internal-to-team** in the user's voice. **External is always draft-only** — the human sends.

## Guardrails (from CLAUDE.md)
- **Draft, never auto-send external** (customer / partner / investor / public). Internal team comms may go in the user's voice.
- **Cite, don't invent** — no unverified facts or figures in a draft; flag the gap instead.
- No confidential figures, PHI/PII, or secrets in the draft or the repo.

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…