Software Engineering Glossary

Progressive Disclosure

Also known as: Lazy loading context On-demand context

Definition

Progressive disclosure is the practice of showing only the information needed for the current step, then revealing more on demand. In AI coding agents it means keeping a short routing hint (a skill name and description) in the context window, and loading the full playbook, reference docs, or scripts only after the agent decides that playbook applies. The same idea shows up in product UI as hiding advanced settings until the user asks for them.

Popular reads

View All
Payment System Design: Ledger, Idempotency, and Settlement
Payment System Design: Ledger, Idempotency, and Settlement Jul 18, 2026
X Algorithm Explained: How the Open Source Recommendation System Works
X Algorithm Explained: How the Open Source Recommendation System Works Jan 22, 2026
Cursor Skills: How to Create and Use Agent Skills
Cursor Skills: How to Create and Use Agent Skills Jun 23, 2026
The Complete HTMX Guide: From Zero to Production
The Complete HTMX Guide: From Zero to Production Dec 22, 2025
Complete Guide to Graph Data Structure: BFS, DFS, Adjacency List vs Matrix
Complete Guide to Graph Data Structure: BFS, DFS, Adjacency List vs Matrix Jan 20, 2026
Transactional Outbox Pattern: Never Lose an Event Again
Transactional Outbox Pattern: Never Lose an Event Again Apr 07, 2026

Key Takeaways

  • You pay tokens for what sits in context. Progressive disclosure keeps unused playbooks cheap.
  • Agent Skills are built around this: metadata first, SKILL.md body second, references/ and scripts/ last.
  • A skill that dumps a 2,000-line style guide into every session is not using progressive disclosure. Split that guide into a linked file.
  • Always-on files such as CLAUDE.md skip this pattern. Put only standing rules there.

How It Works

  1. The host loads a small index of available skills at session start.
  2. The model matches the user request against those short descriptions.
  3. Only the chosen skill’s body enters the conversation. Linked files are read if the instructions say to.
  4. After compaction, hosts like Claude Code may re-attach a truncated copy of recently used skills so the playbook is not lost entirely.

Where It Is Used

  • Claude Code loads skill descriptions for routing, then injects SKILL.md when you type /deploy or when the task matches.
  • Cursor Agent Skills use the same two-stage load so dozens of project skills do not fill the window on every chat.
  • Settings screens that hide power-user options behind an Advanced toggle are the same idea in a GUI.

Related glossary terms