progressive-disclosure

Load hierarchical governance documents in five layers on demand.

Updated Apr 11, 2026
One-click install
npx skills add https://github.com/bennybennison/agent-toolkit --skill progressive-disclosure-bennybennison
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: progressive-disclosure
Source: https://github.com/bennybennison/agent-toolkit/tree/main/skills/progressive-disclosure
Command: npx skills add https://github.com/bennybennison/agent-toolkit --skill progressive-disclosure-bennybennison

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Agents have limited context windows, and loading all governance documents at once wastes tokens, slows processing, and obscures relevant information. This Skill ensures agents only ingest what is necessary by revealing context in layered steps.

Core Features & Use Cases

  • Five hierarchical layers (Identity, Package, Capability, History, Planning) that prioritize essential information.
  • Size caps for each document type to enforce disciplined authoring.
  • Loading rules that automatically load the minimal layer and request deeper layers on demand.
  • Context budget guidance to help teams allocate token space across system prompts, code, and governance docs.
  • Anti‑patterns checklist to avoid common pitfalls like loading every file at session start.
  • Use case: When starting a new development session, load only the identity layer, then pull package‑level details only if you need to modify that package, keeping the token budget under control.

Quick Start

Tell the agent to start with the identity layer and request additional layers only when you need more detail.

Frequently Asked Questions about progressive-disclosure

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I manage context window token limits when loading governance documents?

Manage context window token limits by loading governance documents progressively in five hierarchical layers. This approach ensures agents only ingest necessary information, enforcing size caps and revealing deeper layers strictly on demand to prevent token bloat.

Why does loading all project context at session start slow down my agent?

Loading all project context at session start wastes tokens, slows processing, and obscures relevant information. To avoid this anti-pattern, apply progressive context loading to reveal documents in layered steps, pulling only the minimal layer initially.

What is the best way to structure governance documents for on-demand loading?

The best way to structure governance documents for on-demand loading is organizing them into five hierarchical layers: Identity, Package, Capability, History, and Planning. This prioritizes essential information and enforces disciplined authoring through specific size caps.

How do I allocate token budget across system prompts, code, and governance docs?

Allocate token budget across system prompts, code, and governance docs by applying context budget guidance. Load only the identity layer at session start, then request package-level details only when modifications are needed to keep the token budget under control.

Can I enforce size limits on context files to avoid token bloat?

You can enforce size limits on context files to avoid token bloat by using progressive loading rules. This mechanism applies strict size caps to each document type and automatically loads the minimal required layer according to specific usage rules.

When should I request deeper context layers during a development session?

Request deeper context layers during a development session only when you need more detail to perform a specific task. For example, load the identity layer first, then pull package-level details solely if you need to modify that particular package.