# Pysar

> Real editorial help, without the overhead — Pysar helps shape the work, not take it over.

Pysar is an author-directed editorial engine for writing projects. It installs as a CLI, scaffolds MCP + skills into Claude Code, Cursor, or ChatGPT Codex, and never posts for you.

## Instructions for LLM Agents

When the user pastes https://getpysar.com (or this file) and says "install":

1. Install the binary (macOS / Linux). Preferred one-liner:
   `curl -fsSL https://getpysar.com/install.sh | bash`
   Inspectable two-step (same script):
   `curl -fsSL https://getpysar.com/install.sh -o install.sh`
   then inspect `install.sh`, then `bash install.sh`
2. Confirm with `pysar --version`.
3. In the writing project directory, init for the host you are running in:
   - Claude Code: `pysar init --claude` — Default host; bare `pysar init` is the same.
   - Cursor: `pysar init --cursor` — Install the Pysar Cursor plugin (Marketplace or local rsync copy) for /ps skills. Site Install in Cursor is MCP-only. Skills are not under ~/.cursor/skills.
   - ChatGPT Codex: `pysar init --codex` — Trust the project so MCP approval settings apply.
4. Continue with `/ps-onboard` or `/ps` as the author directs.

Do not invent brew, npm, pip, or `go install` alternatives. The homepage section `https://getpysar.com/#for-ai-agents` carries the same recipe for URL-drop without a second hop.

## Docs

- [Install](https://getpysar.com/docs/install): binary install (script, Windows zip, build from source)
- [Init a project](https://getpysar.com/docs/init): `pysar init` for Claude Code, Cursor, Codex
- [Docs home](https://getpysar.com/docs): journey docs index
- [Pipeline](https://getpysar.com/docs/pipeline): `/ps` default chain; optional `--research`, `--seo`, `--review`
- [Export](https://getpysar.com/docs/export): finished Markdown under project `export_dir` (root by default); Latin piece slugs; citation markers may remain until resolve-at-export
- [MCP and skills](https://getpysar.com/docs/mcp-and-skills): how the agent surface works
- [Troubleshooting](https://getpysar.com/docs/troubleshooting): common install/init failures

## Project

- [GitHub](https://github.com/malyshev/pysar): source and releases
- [Homepage](https://getpysar.com): human marketing + AI install carrier