arrow_back Back to Guides
Architecture Level 04

Workspace Governance Mechanism

Master the PARA governance mechanism. Learn about the Governing Kernel, Invariants, and the Progressive Disclosure principle using governance.md.

Workspace Governance Mechanism

A PARA Workspace is not just a bunch of folders; it is a Governed environment. As you scale, you need tools to ensure the system remains predictable for both you and your AI agents.

📐 The Progressive Disclosure Principle

To prevent context bloat and “attention decay” in AI agents, PARA Workspace v1.4.5+ implements Progressive Disclosure:

  1. Daily Operations: Agents only read a lightweight governance.md rule.
  2. Audit/Planning: Only during these specific workflows is an agent permitted to scan the full Kernel Spec (the constitution).

This ensures the agent stays focused on your task while maintaining high-level compliance.

📜 The governance.md File

Located at the root of your workspace (or in .agent/rules/), this file serves as the “handbook” for agents. It contains:

  • Core Invariants: High-priority rules that must never be broken (e.g., no loose files in root).
  • Workflow Triggers: When to run specific standard procedures like /para-audit.
  • Resource Limits: How to handle tokens and tool calls safely.

Why separate governance?

By separating the Mechanism (this article) from the Tools (Audit and Archive), we ensure that the system architecture is understood before the tools are operated.

→ Next Layer: Workspace Audit → Next Layer: Smart Archive Mechanism → Next Layer: Hybrid 3-File Architecture