chatpharo

Features

This is a user-facing overview of what ChatPharo can do.

Core chat

• Multi-tab chat UI (separate conversations per tab).
• Asynchronous requests (UI stays responsive).
• Cancel in-flight requests.
• Commands: /help, /clear, /reset, /export, /history.

IDE integration

• System Browser integration: context-aware chats for package/class/method.
• Editor actions: “Ask ChatPharo” and “Code ChatPharo” on selected text/method.
• World menu: open Temp Chat, Settings, Setup Wizard, Tutorial, and Wiki.

See integration.md.

Backends (agents)

Choose between local and cloud models through settings.
See agents.md.

Tools (function calling)

When enabled, the model can call tools to inspect the codebase, search methods, and more.
Includes built-in tools, custom tools, and optional MCP tools.
See tools.md.

Skills (domain docs injection)

Skills provide specialized documentation blocks (Spec2, Roassal, Collections, Bloc, etc.) that are injected into prompts.
See skills.md.

Memory (long-term)

ChatPharo can store and retrieve memory items (summaries, preferences, corrections) and inject them into prompts.
See memory.md.

Multivers (prompt chaining)

Multivers runs a configurable chain of steps across agents to improve answer quality for complex tasks.
See multivers.md.

Safety, ethics, sandbox

• Warnings for high-impact changes (prompt edits, tool disabling, message limits, iterations).
• External API warnings for cloud agents.
• Developer gating for ethics settings.
• Optional sandbox restrictions for code execution.

See safety-ethics.md and sandbox.md.

Logging

Optional event logging to a file under the image directory. Helpful for debugging.
See configuration.md.

This site is open source. Improve this page.