# PtahCode

> PtahCode is a persistent, multi-agent AI coding agent desktop app for Windows, macOS, and Linux. It remembers your codebase across sessions, works in parallel across up to nine agents, runs on a schedule, and is reachable from Telegram, Discord, and Slack. It is provider-agnostic: Claude, GitHub Copilot, OpenAI Codex, OpenRouter (200+ models), local Ollama, Kimi K2, and GLM all run through the same app.

## What PtahCode is

PtahCode's desktop app is the product. Unlike autocomplete-class tools (Cursor, GitHub Copilot inline completion), PtahCode is designed around four properties: persistence (it remembers), multi-agent execution (it parallelizes), schedulability (it runs unattended), and reachability (it answers outside the editor).

## The seven differentiators

1. **Persistent memory** — hybrid BM25 + vector search, fused with Reciprocal Rank Fusion, auto-curated across sessions.
2. **Auto-learning skills curator** — extracts, judges, and promotes repeated session trajectories into reusable `SKILL.md` files.
3. **Codebase indexing** — tree-sitter AST parsing and hybrid symbol search across JavaScript, TypeScript, Python, and Go.
4. **Cron scheduler** — SQLite-backed, slot-claimed scheduled agent runs on any cron expression.
5. **Messaging gateways** — trigger and approve agent work from Telegram, Discord, or Slack, including voice input and Discord per-thread sessions.
6. **Sub-agent orchestration** — a main agent fans work out to parallel sub-agents across a three-tier hierarchy, each with its own provider and context.
7. **Workflows and built-in skills library** — pre-built skills for common stacks, extendable via a skills registry.

Bonus: **Orchestra Canvas** — up to 9 concurrent agent sessions in one drag-and-resize grid, each tile independently provider- and model-configured.

## Supported model providers

Claude, GitHub Copilot, OpenAI Codex, OpenRouter (200+ models), Ollama (local models), Kimi K2, GLM.

## Cost

PtahCode is free and open source under the MIT license.

## Canonical links

- Homepage: https://ptah.live
- Download: https://ptah.live/download
- Documentation: https://docs.ptah.live
- GitHub: https://github.com/Hive-Academy/PtahCode