This file is the single source of truth for the multi-agent workflow. Core four roles (required): /manager, /senior-developer, /pair-programmer, /tester. Optional fifth: /slacker (Slack-integrated projects only). All command files reference this spec. If you change the protocol, change it here, and the other commands…
A team mode for crunch throughput: every agent — M included — owns one work item end-to-end in parallel. The serial relay (SD plans → PP reviews → T verifies) is suspended; each owner self-drives the full lifecycle in its own worktree, and the human's PR merge is the final cross-check. M stays the coordinator …
Project-agnostic protocol for sprint retros. M-only writes (standing-authority workflow-artifact pattern). Peers read this file at startup AND on every read-retro-doctrine broadcast — auto-injected by the MCP server busemit handler whenever any caller emits review-closed or retro-open.
Steps every role's -startup.md references. Substitute with your role (manager | senior-developer | pair-programmer | tester | slacker), with the repo root, and with the agent ID you generated.
This file is the canonical token-discipline doctrine for the WOW multi-agent workflow. M reads it once at startup, caches the full text, and pipes it inline as payload.tokendisciplinedoctrine on every story-created. Peers consume it as fresh context per-story-dispatch — bypasses the in-context-prompt-staleness problem.
Tester — write test-stories, exercise features in a per-story worktree, file bugs, verify fixes (via the shared bus).
★not rated 4 todayC27 tokens
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: