Meet the Agents 🦞
← Back to Docs

system/mission.md

Mission

McRitchie Studio is the central task management and orchestration hub for the McRitchie AI agent system. It provides:

  1. Task Pipeline — Create, assign, track, and transition tasks through stages (new → queued → in_progress → done/failed → archived)
  2. Agent Registry — Monitor agent status, skills, activity, and usage
  3. Activity Logging — Track all agent actions for auditability and debugging
  4. Usage Tracking — Monitor API costs, token consumption, and task throughput per agent
  5. Error Capture — Structured error logging with backtrace and context

Core Principle

Agents are autonomous but accountable. Every action is logged, every task is tracked, every size estimate is sealed-bid against actual cost. Each role has KPIs that are damaged by other roles' bad behavior — that's the negotiation surface that keeps quality, throughput, and coherence in healthy tension.

Agents

Personas live at docs/agents/agents/<slug>/{role.md, soul.md}. The DB registry is seeded from db/seeds/02_agents.rb and skills from db/seeds/03_skills.rb.

Leadership

  • Alex — Lead orchestrator (PM). Coordinates agents, manages priorities, makes architectural calls, escalates when Mr. McRitchie's judgment is needed.
  • Avi — Product Owner. Refines tickets, sets po_size (the official planning size), reviews PRs for spec adherence, controls release candidates.

Dev specialists

  • Carl — Backend / Rails. Controllers, models, migrations, jobs, studio-engine internals. Captain of the backend_migration exclusive lane.
  • Shannon — UI. ERB views, Tailwind, Alpine.js, theme system, studio-engine UI primitives.
  • Jasper — Blockchain. turf-vault Anchor program, solana-studio Ruby client, on-chain integration.

Quality + Operations

  • Steffon — QA + Infrastructure. Quality gate on every PR; then Heroku deploys, env vars, CI, observability, recovery protocol.

Domain & support

  • Turf Monster — Sports specialist. Sports data, pick'em games, World Cup props, player analytics.
  • Mack — General worker. Data scraping, processing, API integrations, bulk operations.
  • Mason — Marketing. Brand voice, launch comms, social, funnels, copy.

Agent stack flow

Alex (PM)
  ↔  Avi (PO) ───── refine + assign ────> Devs (Carl, Shannon, Jasper)
                                               │
                                               ▼ open PR
                                          Steffon (QA pass)
                                               │
              ◀───── merge + tag release ──────┘
              │
              ▼
         Steffon (deploy)
              │
              ▼
         Mason (announce)

Off the critical path: Turf Monster (sports domain consults), Mack (data ops, parallel).

System protocols

Three binding protocols shape how the team works. Every soul references them; deviations require Alex's approval.

  • git-protocol.md — worktrees per agent instance, branch convention, PR ownership table, send-back template, 8 git ethics
  • sizing-rubric.md — t-shirt scale, sealed-bid sizing across PM/PO/Dev, accuracy as Avi's primary KPI
  • exclusive-lanes.mdbackend_migration lane, pre-flag vs self-flag paths, Carl's captaincy