Skip to content

Home

Jump to bottom
mastalink edited this page Oct 9, 2026 · 3 revisions

BitCadence User Wiki

Welcome to the BitCadence User Wiki — the complete, step-by-step documentation for installing, configuring, and operating the BitCadence multi-agent orchestration platform. BitCadence Core is source-available (Prosperity 3.0): free for personal use, and businesses license it.

Every guide in this wiki is written from real, hands-on execution. Whether you prefer the Windows native desktop manager, the browser-based Control Panel, or the mco command-line interface, you will find exact numbered instructions, real screenshots, CLI equivalents, and troubleshooting steps.

Table of Contents

Getting Started

  1. Everyday commands (start here): the bitcadence menu and the plain words start, status, ask, approve, fix, connect, pause, helpers and schedule.
  2. Install & First Run — One-click Windows setup, Python environment configuration, and Linux/macOS equivalents.
  3. Desktop Manager & System Tray — Running the local stack with the Windows native control window and notification area tray icon.
  4. Connecting the Console & Demo Mode — Setting up the browser Control Panel, token authorization, and exploring with simulated data.

Day-to-Day Operations

3a. Console Home (what needs you, what is running), Live Score and Helpers: the first page of the console, the read-only live drawing of a Score run, and each helper's light with the Fix it button. Also Schedules, Approvals and Settings, the rest of the console menu. Ask covers asking for something in plain words, and Connect an AI links Claude, Codex, Gemini, Antigravity or Cursor in one tap. 4. Home (Overview) — What needs you, what is running, and the Activity page. 5. Projects Dashboard — Grouping jobs into initiative streams, health monitoring, and coverage analysis. 6. Job Board & Creating Work — Submitting tasks, filtering queues, inspecting job drawers, and managing job lifecycles (retries, cancellations, reassignments). 7. Approvals & Governance — Human-in-the-loop gates, tamper-evident immutable audit trails, and the emergency kill switch.

Flows & Automation

  1. Workflow Files — Saved YAML pipelines. For everyday requests, use Ask for something.
  2. Helpers & Presence — Adding helpers, reading their lights, rotating tokens, and role routing.
  3. Drumline Memory (Shared Context) — Zero-latency collective memory, automatic job distillation, explicit fact recording, and audit-explainable semantic recall.
  4. Drumline Agent Exchange — Helper-to-helper threaded discussion, peer collaboration, and memory promotion.
  5. Autonomous Scores & Conductor — Offline contracts, compile-time validation, policy kernels, and monitoring autonomous score runs.
  6. Jev Routing & Decision Provider — Bounded, deterministic model routing and capability analysis.

System Administration

  1. Settings & Configurations — Gateway connection parameters, visual customizations, LLM provider endpoints, and edition capabilities.
  2. Workers & Background Listening — Connecting AI models (Claude, Codex, Antigravity) via mco listen, mco wake, and MCP.
  3. Scheduling & Recurring Loops — Launchers, interval schedules, cron triggers, and persistent loop execution.
  4. OS Services & Fleet Configuration — Boot persistence via Windows Services / systemd, and declarative worker run modes via fleet.toml.
  5. Sidecars & Autonomous Fleet Daemons — What sidecars are in plain English, installing, starting, stopping, checking, the claude-cio sidecar, elevated admin-pack reconcile, health verification levels, and locked log / duplicate waker troubleshooting.
  6. Complete mco CLI Reference — Recursive reference for all 43 commands and options in the BitCadence command-line tool.

Quality & Usability Audits

In addition to operational walkthroughs, this wiki includes in-depth product audits conducted during hands-on evaluation:

  • Friction Inventory & Flow Surfaces Audit — Complete audit of every user task taking >2 steps, analyzed against non-technical personas (older adults and mobile-only teens), along with an architectural audit of workflow flow surfaces.
  • Bugs & Anomalies Inventory — Detailed reproduction steps for usability and interface bugs identified during testing.

Core Philosophy

  • Mail, Not Function Calls: AI helpers do not directly call each other across arbitrary APIs. They drop auditable mail into bounded dropboxes. Mail can be inspected, gated, retried, and escalated.
  • Local First: Everything runs locally out of the box with zero cloud account, zero external database, and zero external tracking. Local data is safely persisted in SQLite (~/.mco/local.db).
  • Human in the Loop: High-stakes operations pause at cryptographic or token-authenticated gates until authorized by a human approver.
  • Explainable Memory: Drumline collective context provides zero-hallucination, explainable recall scoring across heterogeneous AI providers without external embeddings or cloud dependencies.

Clone this wiki locally