zoom-out

Trigger phrases

Phrases that activate this skill when typed to Claude Code:

• zoom out
• give me the bigger picture
• map the modules in this area

What it does

zoom-out is one of the smallest skills in the Pocock collection — a single instruction, no sub-docs, no procedures. When triggered, the agent stops what it was doing line-by-line and emits a map of the relevant modules and their callers for the area you’re navigating, using the project’s own domain glossary vocabulary rather than generic English.

The skill is marked disable-model-invocation: true in its frontmatter, which means it’s intended to be invoked explicitly rather than auto-fired by the model on a keyword guess. You ask for it; the agent doesn’t decide for you.

When to use it

Reach for it when:

• You’ve been dropped into an unfamiliar section of a large codebase and the line-level view isn’t telling you why anything is shaped the way it is
• You’re about to plan a refactor or feature in territory you haven’t navigated before, and you want a structural read before committing to an approach
• The previous few messages are deep in a single function and you’ve lost the surrounding context
• You’re onboarding to a project and want a forced re-orientation at a domain-vocabulary level rather than a file-tree level

When not to reach for it:

• You already have the structural picture — the output would duplicate what you already hold
• The codebase has no CONTEXT.md or shared glossary; the instruction “use the project’s domain vocabulary” can degrade into generic English without it

Install

The skill is distributed via Pocock’s skills repo. Install via his recommended path (npx skills add or manual copy of the SKILL.md into .claude/skills/zoom-out/) — see the repo README for canonical install instructions.

Once installed, invoke explicitly with one of the trigger phrases above.

What a session looks like

A typical use is a single round-trip: you trigger the skill, the agent produces the map (modules, primary callers, the seams between them), and you read it before deciding the next step. The agent doesn’t continue with the original task automatically — the skill is structural reflection, not a workflow.

The map quality tracks the quality of the project’s shared vocabulary. On a codebase with a maintained CONTEXT.md or glossary, the output reads like a domain map. On a codebase without one, the output drops to a generic-English module list — useful but lower-leverage.

Receipts

TODO — to be filled in from a real session. When the skill is triggered in production use, capture: which area of the codebase you were in, what the map surfaced that the line-level view hadn’t, and whether the project’s domain glossary made the output domain-specific or generic.

Source and attribution

Originally written by Matt Pocock. The canonical SKILL.md lives in the engineering/zoom-out folder of his public skills repository.

License: MIT. You can install, adapt, and redistribute the skill, with attribution preserved.

This page documents the skill from a practitioner’s perspective. For the formal spec and any updates, defer to the source repo.