Skip to content
Agentmart
Submit a tool

dont-anger-mario

Prepare concise, high-signal issues or contribution proposals for the pi repository without annoying maintainers.

Skill by dannote

IndexedMIT★ 57
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/dannote-dont-anger-mario

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

Prepare concise, high-signal issues or contribution proposals for the pi repository without annoying maintainers. Use when filing issues/PR proposals for earendil-works/pi or mentioning Mario/pi tracker etiquette.

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 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/dont-anger-mario

What's inside

Skill name
dont-anger-mario
Description
Prepare concise, high-signal issues or contribution proposals for the pi repository without annoying maintainers. Use when filing issues/PR proposals for earendil-works/pi or mentioning Mario/pi tracker etiquette.
License
MIT
Instructions
109 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Don't Anger Mario

Use this skill before opening an issue or contribution proposal in `earendil-works/pi`.

## Rules

- Read `CONTRIBUTING.md` first.
- Use an official issue template; blank issues are disabled.
- Keep it short. If it does not fit on one screen, it is too long.
- Write in the user's own voice, not corporate/AI slop.
- Be concrete: what, why, and optionally how.
- Do not submit automatically.
- Before submitting, ask the user to confirm they carefully read the exact final draft and explicitly approve posting it.
- If proposing a change, use **Contribution Proposal** rather than Bug Report.
- If reporting a bug, include minimal repro and version.
- If opening a PR, only proceed after maintainer approval (`lgtm` for PRs; `lgtmi` is issues only).
- Do not send agent-only PRs. If the user cannot explain the change without the agent, stop.

## Safe workflow

1. Inspect templates and contributing docs:

```bash
gh api repos/earendil-works/pi/contents/.github/ISSUE_TEMPLATE --jq '.[].name'
gh api repos/earendil-works/pi/contents/CONTRIBUTING.md --jq .content | base64 --decode
```

2. Draft the issue body in `/tmp`, never inline long Markdown through shell quoting:

```bash
cat > /tmp/pi-issue.md <<'EOF'
### What do you want to change?

...

### Why?

...

### How? (optional)

...
EOF
```

3. Show the exact final title and body to the user.

4. Ask for explicit approval with this wording:

```text
Please read the draft carefully. Do you explicitly approve me posting this issue/proposal as written?
```

Do not proceed on vague approval like "ok" if the user has not seen the exact final draft.

5. Create only after explicit approval:

```bash
gh issue create \
  --repo earendil-works/pi \
  --template contribution.yml \
  --title "Short concrete title" \
  --body-file /tmp/pi-issue.md
```

## Extra caution from prior tracker behavior

Known maintainer reactions that shape this workflow:

- Auto-close gate is intentional; maintainers review closed issues later.
- GitHub Actions outages can let the issue/PR gate fail and flood the repo.
- Maintainers dislike agent-only PRs because they take time to review and are often broken.
- "Clanker misunderstood the code" is a real failure mode; verify the claim before drafting.
- A label/comment like `possibly-openclaw-clanker` is a warning sign: shorten, verify, and make the human reasoning obvious.

Useful receipts:

- https://github.com/earendil-works/pi/issues/3589 — "very tired of clankers"
- https://github.com/earendil-works/pi/issues/5259 — "your clanker misunderstood"
- https://github.com/earendil-works/pi/pull/2666 — "You and your clanker are banned"
- https://github.com/earendil-works/pi/pull/5296 — "please do not send more agent only PRs"

## Good contribution proposal shape

```md
### What do you want to change?

One concrete change.

### Why?

The user-visible problem and why it matters.

### How? (optional)

Small implementation sketch, if useful. Avoid over-design.
```

## Tone

Good:

> Prompt templates support `$1` and `$ARGUMENTS`, but optional arguments cannot have defaults. I want `/next` and `/next 3` to expand cleanly without asking the model to interpret an empty argument.

Bad:

> Implement advanced Bash-compatible templating with all parameter expansion semantics, conditionals, filters, and expression evaluation.

From dannote/dot-pi, MIT.

More from dannote/dot-pi

Signals

Stars
57
Forks
4
Open issues
0
Last commit (this folder)
Jun 8, 2026 (4 months ago)
Last release
Aug 30, 2026 (2 months ago)

From GitHub (fetched Oct 24, 2026).

Details

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

More from dannote

Cite this page

dont-anger-mario skill. Agent Mart. https://agentmart.dev/skills/dannote-dont-anger-mario (last updated 2026-10-24)