Member-only story
Check what Muse Code reads from Claude Code before you pick the contributor tier
I use Muse Spark 1.3 on it and recommend it. By default, Muse Code can also pick up your CLAUDE.md and Claude Code skills.
The model I run in Muse Code has -contributor on the end of its name.
That suffix is the entire deal. On Meta’s pricing page, standard Muse Spark 1.3 costs $1.25 per million input tokens and $4.25 per million output, and your prompts and completions are not used to train Meta models.
The contributor version is $0.10 and $0.20, “in exchange for permission to use your prompts and completions to train future Meta models.”
I picked the second one. On real work, on my own project, and I’d tell a friend to try it. I’m not unusual, either: Meta AI chief Alexandr Wang told Axios that a “meaningful double digit” percentage of coders choose the contributor option.
I also run Claude Code most days, and I’ve written Claude Code skills of my own, so the sentence in Muse Code’s docs I keep coming back to is about skills: Muse Code “also discovers ~/.claude/skills” by default. More on that in a minute.
What using it was actually like
Muse Spark 1.3 came out on 2 September, and Muse Code is Meta’s own coding agent for it, a terminal…
