Workflow diagrams

Mermaid renders on GitHub, in Obsidian, and on the docs site.

From first message to a working company

Nobody is asked to choose a command. Crew mode is the default offer after /multica-ops:import when no conductor is standing — someone who just moved their backlog has already decided what the work is; a workspace that already has a conductor keeps it. Adding a conductor later is an upgrade, not a redo.

The four routes, side by side

Two seats of Mops

One feature through the conveyor

A stage is a barrier, not a queue: everything genuinely independent goes on the same stage and runs concurrently — the numbers order dependencies, not tasks. And width is only real on a github_repo project; a local_directory serializes everything regardless.

Escalation & control

Session limits — detect and recover

Getting current — what /upgrade actually walks

Two things this picture exists to prevent: new bytes without a migration (half the company on one version, half on another), and a CLI replaced under a running agent, which produces failures that look like the agent's fault.

Something went wrong — whose defect is it?

The person who just hit it should not have to classify it. The door decides, from evidence.

/multica-ops:report is the door, and a sentence reaches the same place. The file is written before anything is missing — a field it cannot know is marked unknown and the offer to fill it comes after, because a missing field is not a reason to withhold the artefact. Measured: scenario 23 caught the first version stopping to ask, leaving the report as chat text.

The skill lifecycle — gates, not ceremony

The loop back to Screen on upgrade is the point most setups miss: a version you vetted is not the version you are about to install.

A catch climbs, and a contradiction stops

Two rules that changed how this skill maintains itself, drawn because both are easy to nod at and hard to follow.

Three attempts bound failure; the second diagram bounds contradiction — the worse state, because every run inside it looks like a success. Both are listed by name in the prose-only table with the exact reason neither is a gate today: issue runs records that a run finished, not what it concluded, and nothing compares a promotion's citation against its note's date.