Thoth

A coding agent for the model you choose. Point Thoth at your own hardware (Ollama, llama.cpp, vLLM) or at an api you pay for (Anthropic, OpenAI, Google, OpenRouter) and it explores your codebase, edits files, runs builds and tests, and searches the web when it needs documentation. Every action is shown on screen and destructive ones wait for your approval.

Highlights

• Your model, your machine. Auto-detects Ollama and uses its native API to control the context window itself, so agentic prompts don't get silently truncated. Anthropic gets its own transport, with prompt caching. Everything else speaks the OpenAI api.
• Profiles. Keep local and hosted setups side by side, switch with one key, and see what a session has cost while it runs. thoth config.
• Safe by construction. The model cannot edit a file it hasn't read, or overwrite one it hasn't read completely. These rules live in Rust, not in the prompt. Full diffs and full command lines are always displayed, even after "always allow".
• Editor aware. With the VS Code extension, Thoth knows your active file, selected lines and the Problems panel, and the model can re-check diagnostics after every edit.
• Built for small context windows. Live token meter, automatic compaction, per-project memory and session recaps keep long tasks running on 16-32k tokens.
• No account needed. Web search goes through DuckDuckGo, with no key.

Install

Linux / macOS:

curl -fsSL https://raw.githubusercontent.com/thoth-coder/thoth/main/scripts/install.sh | sh

Windows (PowerShell):

irm https://raw.githubusercontent.com/thoth-coder/thoth/main/scripts/install.ps1 | iex

Or build from source: cargo build --release (-> target/release/thoth). Update later with thoth upgrade.

Quick start

Local:

ollama pull qwen3:8b             # any tool-calling model
thoth                            # or: thoth -p "one-shot prompt"

Hosted: thoth config and fill in a profile, e.g. base url https://api.anthropic.com/v1 with your key and claude-sonnet-4-5. See Configuration for OpenAI, Google and the rest.

Documentation

Getting started	Ollama, llama.cpp, choosing a model
Usage	keys, commands, memory, editor integration
Tools & guardrails	what the model can do and how it is contained
Configuration	profiles, providers, cost, context window

Contributing

Bug reports and pull requests are welcome. See CONTRIBUTING.md for the ground rules and SECURITY.md for the threat model.

License

Dual-licensed under Apache-2.0 or MIT, at your option.