π Website
https://github.com/solvyxtech/molt
π¦ Ready to be included in x-cmd pkg ?
NO
π₯ How to install
Desktop app and CLI builds (macOS / Windows / Linux): https://github.com/solvyxtech/molt/releases/latest
From source:
git clone https://github.com/solvyxtech/molt
cd molt
npm install
npm run build
npm start # terminal UI
(The @solvyx/molt npm package has been withdrawn, so the earlier npm i -g / npx lines no longer work.)
π TLDR
# interactive TUI β pick provider, model, then work a project
molt
# headless: exits non-zero if the project's .molt/done.yml bar is not met
molt run "fix the failing test" --yes
# re-run verification only
molt prove
# recompute the receipt hash chain
molt verify
π Highlights
- Harness, not another chatbot. molt is an open-source coding agent for terminal and desktop that refuses to accept "done" until checks in
.molt/done.yml pass against the real on-disk state.
- Accepts and refusals both write a receipt; writes get before/after hashes; journal is hash-chained (
molt verify).
- Runs OpenAI-compatible endpoints (Ollama, vLLM, etc.) or Anthropic.
- Current release v0.2.2 (CLI binary name
molt; desktop app is branded Maat Agent). License Apache-2.0. Studio: https://solvyx.xyz/work/molt
- Please do not confuse with unrelated npm packages named
molt / molt-cli.
Stories: built so a model cannot high-five itself for unfinished work β acceptance lives outside the model.
Disadvantages / honest limits: needs a reachable model endpoint and a project with a usable done-bar; young project (stars still early).
π Website
https://github.com/solvyxtech/molt
π¦ Ready to be included in x-cmd pkg ?
NO
π₯ How to install
Desktop app and CLI builds (macOS / Windows / Linux): https://github.com/solvyxtech/molt/releases/latest
From source:
(The
@solvyx/moltnpm package has been withdrawn, so the earliernpm i -g/npxlines no longer work.)π TLDR
π Highlights
.molt/done.ymlpass against the real on-disk state.molt verify).molt; desktop app is branded Maat Agent). License Apache-2.0. Studio: https://solvyx.xyz/work/moltmolt/molt-cli.Stories: built so a model cannot high-five itself for unfinished work β acceptance lives outside the model.
Disadvantages / honest limits: needs a reachable model endpoint and a project with a usable done-bar; young project (stars still early).