Voltron

Voltron is formed, the way the lions join into one robot: in every status line and summary the verb is form (Turkish oluşturmak: "Voltron oluşturuyoruz", "Voltron oluşturuldu"). The Elder splits the job at hand into parts, each lion builds one part at the same time, and the Elder joins the parts into one working whole. Read lions first, for who the lions are, choosing models, the plan format, running and reading results; in a repository checkout it is ../cogitors/references/lions.md.

Form Voltron

1. Split. Break the current job into independent parts, one per lion, at most five; a small job uses fewer lions. Give each part its own owns paths so no two lions change the same file. When parts meet at an interface, write that shared scaffold yourself first (types, interfaces, stubs, the data contract) in the working tree; every lion starts from the working tree as it is.
2. Match lions to parts. Choose each lion's model and effort from cogitor.py models by what its part demands.
3. Announce. Say "Let's form Voltron!" in the user's language (Turkish: "Voltron oluşturuyoruz!"), then the plan: lion · model · effort → part (paths). Start right away unless the user asked to see the plan first.
4. Run a parts plan with --access edit, or review when every part only reads (for example a review split by area).
5. Join. Skip lions whose status is not success or whose changed is false, and decide what to keep of any outside_owns change. Apply the diffs one by one, resolve conflicts, then run the project's own checks and fix what the joining broke. This step is done when every part is applied or set aside with a reason and the checks pass, or their failures are named.
6. Report to the user in their language: what each lion built, what you changed while joining, the check results, and any part left undone. Then clean the run.