Skip to content
Agentmart
Submit a tool

draft

Save Claude's previous chat response (or the answer to a new question) as a plain markdown file under .notes/drafts/ so the user can annotate it in the editor…

Skill by davidsvaughn

IndexedMIT
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/davidsvaughn-draft

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Save Claude's previous chat response (or the answer to a new question) as a plain markdown file under .notes/drafts/ so the user can annotate it in the editor with CriticMarkup and send it back.

Use cases: documents, productivity

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/draft

What's inside

Skill name
draft
Description
Save Claude's previous chat response (or the answer to a new question) as a plain markdown file under .notes/drafts/ so the user can annotate it in the editor with CriticMarkup and send it back.
License
MIT
Instructions
58 lines in SKILL.md · view on GitHub

Files (2)

  • SKILL.md
  • criticmarkup.md · references/

SKILL.md (first lines)

# /draft — a response as an annotatable file

Timestamp for the filename: !`date +%Y%m%d-%H%M%S`
Date for the header: !`date '+%Y-%m-%d %H:%M'`
Session id: ${CLAUDE_SESSION_ID}
(If the lines above show a command or a variable name instead of values, this harness does not fill
them in: run `date +%Y%m%d-%H%M%S` and `date '+%Y-%m-%d %H:%M'` yourself, and leave the session id empty.)

Arguments: the text the user typed after the skill name, which the harness appends after these
instructions (Claude Code as `ARGUMENTS: ...`, pi/OMP as `User: ...`); none appended means no arguments.
Paths in this file are relative to this skill's directory, which is stated where the skill was loaded.

## What to do

**If the arguments line above is empty:** write your *previous* chat response — the last full
assistant message before this command — to a file, **verbatim**. Do not summarize, tidy, reorder,
shorten, renumber, or add line numbers or section numbers. Keep every heading, table, list, code
block and link exactly as written, including any closing question or offer.

**If arguments were given:** treat them as a question or request and write your full answer to the
file instead of to chat, in the form you would have used in chat. Same rules: plain markdown, no
line numbers.

## File

- Directory: `.notes/drafts/` in the current working directory. Create it if missing. Keep it out
  of git: if `git check-ignore -q .notes/x` fails, tell the user once that `.notes/` belongs in
  their global gitignore (`~/.gitignore` or `core.excludesFile`), and do not add the file to git.
- Name: `<timestamp>-<slug>.md`, where `<slug>` is 2–5 lowercase hyphenated words naming the topic.
  The `yyyymmdd-hhmmss` prefix keeps alphabetical order chronological.
- Content: the front-matter header below, then the body, nothing else — no added title, no wrapper,
  no line numbers. The body is the verbatim previous response (bare `/draft`) or the full answer
  (`/draft <question>`).

Header (fill from the values at the top of this file):

```
---
date: <date for the header>
session: <session id>
kind: captured-response | file-answer
prompt: <the user message this body answers, trimmed to one line — for bare /draft, the message that preceded your previous response; for /draft <question>, the question>
---
```

## After writing

Reply in chat with one line and nothing else:

`Draft saved: .notes/drafts/<file>.md`

The user will annotate the file with CriticMarkup — `{==text==}{>>you|date: comment<<}`,
`{>>you|date: comment<<}`, `{++insert++}`, `{--delete--}`, `{~~old~>new~~}` — and come back with
"address the annotations in <path>". Handle that as `references/criticmarkup.md` says (read it then if it is not already in context): inventory every annotation before acting, work
out the scope of each, act on its intent, respond in chat rather than in the file, and never edit,
move or remove the user's annotations.

From davidsvaughn/skills, MIT.

Signals

Stars
0
Forks
0
Open issues
0
Last commit (this folder)
Sep 30, 2026 (29 days ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from davidsvaughn

Cite this page

draft skill. Agent Mart. https://agentmart.dev/skills/davidsvaughn-draft (last updated 2026-10-24)