# pi-prompting

Internal prompt guidance for Pi task runners using DeepSeek models. It helps structure coding, debugging, and repository review through strict XML contracts.

- skill · `type=skill` · Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, OpenCode · 2900 installs · ★ 92
- Author: Agents365-ai
- License: not stated · policy=link-only · description only; the file stays with the author
- Source: https://github.com/Agents365-ai/365-skills/tree/HEAD/plugins/pi-plugin-cc/skills/pi-prompting/SKILL.md
- Updated: 2026-10-16
- HTML: https://affpapa.org/en/nevibecode/skills/pi-prompting

## What it does

This skill guides `pi:pi-companion-forwarder` when it delegates work to Pi. By default, the plugin targets DeepSeek V4 models: Flash for general work and Pro for code reviews. Instead of loose instructions, prompts are organized into explicit XML blocks like `<task>`, `<output_contract>`, and `<action_safety>`. Numbered checklists take priority over plain paragraphs.

For coding and debugging, prompts include a `<completeness_contract>` requiring passing tests and clean lint runs. When `pi-subagents` is present, it adds an `<available_pi_subagents>` block to route tasks between workers, planners, and scouts. Built-in tools like `/pi:review` and `/pi:adversarial-review` already carry predefined review instructions, so manual prompt drafting is unnecessary for those tasks.

Raising reasoning with `--effort high` by default is counterproductive. DeepSeek Flash gets no benefit from it, while DeepSeek Pro takes longer to answer. Tightening the prompt checklist works better.

## Install

### Claude Code (native platform)

1. **Skills CLI (Vercel)**
```bash
npx skills add Agents365-ai/365-skills --skill pi-prompting -a claude-code
```

2. **Manually: put the files in place**
- File: `~/.claude/skills/pi-prompting/SKILL.md` — Copy the whole skill folder; for one project use .claude/skills/pi-prompting/

### Codex

1. **Skills CLI (Vercel)**
```bash
npx skills add Agents365-ai/365-skills --skill pi-prompting -a codex
```

2. **Manually: put the files in place**
- File: `.agents/skills/pi-prompting/SKILL.md` — For all projects use ~/.agents/skills/

### Cursor

1. **Skills CLI (Vercel)**
```bash
npx skills add Agents365-ai/365-skills --skill pi-prompting -a cursor
```

2. **Manually: put the files in place**
- File: `.cursor/skills/pi-prompting/SKILL.md` — Cursor also reads .agents/skills/ and .claude/skills/

### GitHub Copilot

1. **GitHub CLI**
```bash
gh skill install Agents365-ai/365-skills pi-prompting
```

2. **Manually: put the files in place**
- File: `.github/skills/pi-prompting/SKILL.md`

### Gemini CLI

1. **Skills CLI (Vercel)**
```bash
npx skills add Agents365-ai/365-skills --skill pi-prompting -a gemini-cli
```

2. **Platform CLI**
```bash
gemini skills install https://github.com/Agents365-ai/365-skills --path plugins/pi-plugin-cc/skills/pi-prompting
```

### Windsurf (Devin Desktop)

**Manually: put the files in place**
- File: `.devin/skills/pi-prompting/SKILL.md` — The legacy .windsurf/skills/ path still works

### OpenCode

**Skills CLI (Vercel)**
```bash
npx skills add Agents365-ai/365-skills --skill pi-prompting -a opencode
```

### Claude.ai

**Claude.ai**
- Zip the folder and upload it in Settings → Capabilities → Skills

Where a file is given instead of a command, copy the files from the source to that path: https://github.com/Agents365-ai/365-skills/tree/HEAD/plugins/pi-plugin-cc/skills/pi-prompting/SKILL.md

## When to use it
- structure Pi prompts with `<task>` and `<action_safety>` XML tags
- restrict writable directories and files during automated edits
- enforce definition-of-done criteria with tests and linter checks
- route subtasks across specialized workers with `pi-subagents`

## Example prompts
- Draft a task prompt for Pi to fix a bug in `src/auth.ts` following pi-prompting rules
- Prepare a prompt for Pi with an `<action_safety>` block for safe refactoring
- Compose a follow-up instruction for `task --resume-last` with only the delta

## Similar
- [gpt-5-4-prompting](https://affpapa.org/en/nevibecode/skills/gpt-5-4-prompting.md) — skill · 728 installs
- [langchain-dev-guide](https://affpapa.org/en/nevibecode/skills/langchain-dev-guide.md) — skill · 836 installs
- [fal-prompting](https://affpapa.org/en/nevibecode/skills/fal-prompting.md) — skill · 510 installs
- [cli-anything-novita](https://affpapa.org/en/nevibecode/skills/cli-anything-novita.md) — skill · 462 installs
- [claude-switch-models-setup](https://affpapa.org/en/nevibecode/skills/claude-switch-models-setup.md) — skill · 344 installs
- [read](https://affpapa.org/en/nevibecode/skills/read.md) — skill · 14000 installs

From the same set: [agents365-ai/365-skills](https://affpapa.org/en/nevibecode/by/agents365-ai/365-skills.md) — 21 cards

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md