use Multiple models for bunch of tasks - code-review, verifying task is done, adhd planning or just questions.
currently it can use Claude Code's subagents + Codex + OpenCode + Gemini + Headless Claude Code with openrouter key if you provide one. use whatever you want but good base setup - claude code + codex, other are optional.
it can use all of those models but it doesnt mean its gonna be better because theyall still llms. just use whatever you have or you want (or everything at once idc yolo)
soo basically its what the name implies - several models instead of one:
/multi:code-review- code-review with Claude's sub-agents, Codex request (via headless exec) and OpenCode's headless request + ponytail lens for overengineering. just read ponytail plugin desc, its kinda cool/multi:check-if-done- the "no bro is it REALLY done" one. a model that just wrote the code is the worst judge of whether it works, so this asks models that didn't write it, and refuses to call anything done without actually running a command that proves it. should fix "looks done but not really done sorry lmao" cases/multi:ask- just ask everyone the same thing and see three answers - u'll get three answers, no merging, no judjign/multi:adhd- summons every model with adhd skill. like mega-plan js read skill lmao its cool/multi:setup- tells you what this plugin is about and how to connect everything and etc
install
claude plugin marketplace add szarkans/multi
claude plugin install multi@szkills
or
npx skills add szarkans/multi
or
git clone https://github.com/szarkans/multi ~/.claude/skills/multithen restart claude code and /multi:setup
im tired of ai-slop-saas-b2b-skill readme's. plain user-readable text is better, no garbage noise, only what you actually need.
like... is there something i missed? i can guide you throuugh every script but do you really want it?
I mean... It works for me? Code-reviews got so much better and check-if-done is OP. i like it.
MIT © Sergei Shatrov