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/rizukirr-plain
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.
About
Use cases: writing
Reviews
Write a reviewNo 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 and code. 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/plain
What's inside
- Skill name
plain- Description
- Use before writing any text, in chat or into a file. Thirteen rules that hold in every string you emit, including code comments, commit messages, PR bodies and docs. No em dash, no semicolon, no hard wrapping inside a paragraph, no curly quotes, no decorative emoji, no title case in headings, no bold label lists, no heading echo, no fragment runs, no padded triads, no rejected straw options, no unraised objections, no writing about the previous version. Stays on after.
- License
- MIT
- Instructions
- 67 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# plain Thirteen rules about how text is typed and shaped. They hold everywhere, with no exception for artifacts. ## Persistence Invoke once, then active every response. No drift back after many turns. Still active if unsure. Off only on "stop plain" or "normal mode". ## The rules ### Typography 1. **No em dash.** No en dash either, no spaced dash such as ` - ` used as one, no double hyphen. Not in prose, not in a table cell, not in a heading, not in a frontmatter description. Use a comma, a colon, or a full stop. If the sentence needs the pause an em dash gives it, it is two sentences. 2. **No semicolon.** Split into two sentences. A list whose items contain commas is the one place a semicolon earns its keep, and even there a bulleted list is better. 3. **No hard wrapping inside a paragraph.** One paragraph is one line, however long. Wrap at the paragraph, never at a column. In code comments, follow the project linter instead, because a formatter that reflows comments will fight this rule and the formatter wins. 4. **No curly quotes.** Straight quotes only, for both double and single marks, and a straight apostrophe. A curly mark is what a word processor produces, not what a terminal, a diff or a code file wants. 5. **No decorative emoji.** Not at the start of a heading, not at the start of a list item, not as a status marker in prose. An emoji the user asked for is not decoration and stays. 6. **Sentence case in headings.** Capitalise the first word and any proper noun. Not every main word. ### Shape 7. **No bold label lists.** A list where every item opens with a bold phrase and a colon is a table pretending to be prose. Write the prose, or write an actual table. 8. **No heading echo.** The sentence after a heading must not restate the heading. Delete it and start with the content. 9. **No fragment runs.** One short sentence for emphasis is fine. Three or more clipped fragments in a row is a drum roll, and it reads as performance rather than information. 10. **No padded triads.** Use the number of items the meaning has. Two is a fine list. Four is a fine list. Reaching for a third item because three sounds complete is how a real point gets a filler sibling. ### Drafting residue 11. **No rejected straw options.** Do not raise an option nobody proposed in order to dismiss it in the same clause. State the constraint directly. 12. **No unraised objections.** Do not answer a challenge the text never made. "To be clear, this is not about X" belongs in a reply to someone who said X. 13. **No writing about the previous version.** Describe what the code does now. Prior behaviour belongs in a changelog, a migration note, a release note, a root cause, or a commit message. ## Not a tell by itself Rules 7 through 13 are judgment calls, and each has a legitimate form that must survive. Three items are fine when there are three things. The tell is padding to reach three. One short sentence for emphasis is fine. The tell is a run of them. A heading followed by a definition is fine. The tell is a restatement. An alternative is fine when it is weighed and its trade-offs are given. The tell is one raised and dropped in the same clause. `brainstorm` requires two or three approaches with trade-offs and an Alternatives considered section, and rule 11 never overrides that. Prior behaviour is fine in a changelog, a migration note, a release note, a root cause or a commit message. The tell is it appearing in a description of current behaviour. `debug` produces root causes and `exec` writes commit messages, and rule 13 never overrides either. ## Where this applies Every string that leaves the session. Prose to the user, code comments, docstrings, commit messages, PR bodies, markdown docs, specs and plans. The one exception is text a skill requires reproduced verbatim. When an instruction says to output something word for word, output it word for word, banned characters included, and the rule yields. ## Boundaries How text is typed and shaped, not how much of it there is. `terse` covers volume, `lazy` covers code.
From rizukirr/vibekit, MIT.
More from rizukirr/vibekit
- 0.644score
quick
rizukirr
Use when the user types /vibekit:quick: writes the change immediately under lazy's ladder, no spec, no plan, no subagents.
SkillAFreshness: A★ 20Claude Code +9 - 0.644score
exec
rizukirr
Use when a plan is approved and implementation has not started: dispatches one fresh subagent per task, runs each task's verify clause, and routes failures…
SkillAFreshness: A★ 20Claude Code +9 - 0.644score
using-vibekit
rizukirr
Use when starting any conversation: establishes the auto-trigger discipline so guardrail skills fire instead of being silently skipped.
SkillAFreshness: A★ 20Claude Code +9 - 0.644score
debug
rizukirr
Use when a check fails: a red test, a broken build, a failed clause, or a bug you can point at.
SkillAFreshness: A★ 20Claude Code +9 - 0.644score
plan
rizukirr
Use when a spec is approved and implementation has not started: turns it into a task-by-task plan with exact paths and checkable verification.
SkillAFreshness: A★ 20Claude Code +9 - 0.641score
brainstorm
rizukirr
Use before any creative or implementation work: features, components, behavior changes.
SkillAFreshness: A★ 20Claude Code +9 - 0.641score
lazy
rizukirr
Use at the start of any coding work: writing, adding, refactoring, fixing, designing.
SkillAFreshness: A★ 20Claude Code +9 - 0.641score
terse
rizukirr
Use at the start of every session.
SkillAFreshness: A★ 20Claude Code +9 - 0.641score
verify
rizukirr
Use before claiming a change is done, fixed or passing: checks the whole change against its spec, runs the checks no single task could, hunts for the bugs the…
SkillAFreshness: A★ 20Claude Code +9 - 0.633score
vibe
rizukirr
Run a short intent through the pipeline.
SkillAFreshness: A★ 20Claude Code +9
Signals
- Stars
- 20
- Forks
- 1
- Open issues
- 0
- Last commit (this folder)
- Sep 5, 2026 (1 month ago)
- Last release
- Oct 10, 2026 (19 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/rizukirr/vibekit
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from rizukirr
- 0.644score
quick
rizukirr
Use when the user types /vibekit:quick: writes the change immediately under lazy's ladder, no spec, no plan, no subagents.
SkillAFreshness: A★ 20Claude Code +9 - 0.644score
exec
rizukirr
Use when a plan is approved and implementation has not started: dispatches one fresh subagent per task, runs each task's verify clause, and routes failures…
SkillAFreshness: A★ 20Claude Code +9 - 0.644score
using-vibekit
rizukirr
Use when starting any conversation: establishes the auto-trigger discipline so guardrail skills fire instead of being silently skipped.
SkillAFreshness: A★ 20Claude Code +9 - 0.644score
debug
rizukirr
Use when a check fails: a red test, a broken build, a failed clause, or a bug you can point at.
SkillAFreshness: A★ 20Claude Code +9 - 0.644score
plan
rizukirr
Use when a spec is approved and implementation has not started: turns it into a task-by-task plan with exact paths and checkable verification.
SkillAFreshness: A★ 20Claude Code +9 - 0.641score
brainstorm
rizukirr
Use before any creative or implementation work: features, components, behavior changes.
SkillAFreshness: A★ 20Claude Code +9
Cite this page
plain skill. Agent Mart. https://agentmart.dev/skills/rizukirr-plain (last updated 2026-10-24)