Short answer: Nimbalyst reviews agent work so you can ship the code. Stet reviews agent writing so you can sign the document.
Nimbalyst is a cockpit for coding agents. Run Claude Code and Codex in parallel sessions, watch their diffs, approve changes, track tasks on a board. If you're building software with agents, it's a genuinely good way to stay in control of a lot of moving work, and it's open source. We mean the comparison as a compliment.
In Nimbalyst, documents are scaffolding. Specs, task lists, and notes exist on the way to the real deliverable, which is working software. Review exists so the code merges safely.
In Stet, the document is the deliverable. The spec that ships, the brief the agent acts on, the SOW that gets signed. Review exists because your name is on the words.
That difference shows up in scope. Nimbalyst reviews the sessions it runs: agents launched inside its workspace. Stet watches the folder itself, so a change from any agent, any terminal, any script arrives as a proposal, whether Stet started it or not. And Stet keeps the record with the file: who wrote each change, and that you approved it, for the life of the document.
Plenty of people are both builders and authors in the same week. Run your coding agents in Nimbalyst. Keep the documents that carry your name in a folder Stet watches. Same plain markdown, no conflict between them.
Part of Where Stet fits, the map of the markdown neighborhood.