Agents

Reference, audit, and diagnostic agents

Agents run in isolated context. Use reference agents for focused API lookups, the auditor for code review, and the diagnostics agent for debugging broken text.

Apple Text ships 6 agents: 4 domain reference agents, 1 code auditor and 1 diagnostic.

Domain Reference Agents

Domain reference agents load relevant skills on demand and answer specific questions in isolated context. The router skill and /apple-text:ask command launch these automatically when a reference lookup is needed.

editor-reference

Look up editor feature APIs — Writing Tools, text interaction, text input, undo/redo, find/replace, pasteboard, spelling, drag-and-drop, accessibility, and Dynamic Type.

Best for

• specific API lookups and reference questions
• implementation guidance for a known subsystem
• getting a focused answer without flooding the main conversation

You get

• a focused answer extracted from the full reference
• exact API signatures and code examples
• isolated context that keeps the main conversation clean

platform-reference

Look up SwiftUI bridging, UIViewRepresentable wrappers, TextEditor iOS 26+, AppKit vs UIKit differences, TextKit 1 vs 2 selection, Core Text, Foundation text utilities, and parsing.

Best for

• specific API lookups and reference questions
• implementation guidance for a known subsystem
• getting a focused answer without flooding the main conversation

You get

• a focused answer extracted from the full reference
• exact API signatures and code examples
• isolated context that keeps the main conversation clean

rich-text-reference

Look up attributed string APIs, text formatting attributes, colors, Markdown rendering, text attachments, line breaking, and bidirectional text.

Best for

• specific API lookups and reference questions
• implementation guidance for a known subsystem
• getting a focused answer without flooding the main conversation

You get

• a focused answer extracted from the full reference
• exact API signatures and code examples
• isolated context that keeps the main conversation clean

textkit-reference

Look up TextKit 1/2 APIs, layout mechanics, viewport rendering, text measurement, exclusion paths, fallback triggers, and text storage patterns.

Best for

• specific API lookups and reference questions
• implementation guidance for a known subsystem
• getting a focused answer without flooding the main conversation

You get

• a focused answer extracted from the full reference
• exact API signatures and code examples
• isolated context that keeps the main conversation clean

Specialist Agents

textkit-auditor

Use this agent when the user mentions TextKit review, text view code review, or asks to scan for TextKit anti-patterns. Automatically scans Swift/Objective-C code for TextKit issues — detects TextKit 1 fallback triggers, deprecated glyph APIs, missing editing lifecycle calls, unsafe text storage patterns, and Writing Tools compatibility problems.

Best for

• findings-first reviews of editor code
• scanning a codebase for TextKit anti-patterns
• focused audits where file references and fix directions matter

You get

• findings grouped by severity
• file references for each issue
• one concrete fix direction per finding

Audit focus

• TextKit 1 fallback triggers
• deprecated glyph and layout APIs
• missing editing lifecycle calls
• unsafe text storage patterns
• Writing Tools compatibility problems

textkit-diagnostics

Use this agent when the user describes a broken text symptom — stale layout, editing crashes, TextKit 1 fallback, Writing Tools not working, rendering artifacts, typing lag, or text content loss. Autonomously reads the user’s code, follows a diagnostic decision tree, and returns root cause + fix.

Best for

• findings-first reviews of editor code
• scanning a codebase for TextKit anti-patterns
• focused audits where file references and fix directions matter

You get

• findings grouped by severity
• file references for each issue
• one concrete fix direction per finding

Audit focus

• TextKit 1 fallback triggers
• deprecated glyph and layout APIs
• missing editing lifecycle calls
• unsafe text storage patterns
• Writing Tools compatibility problems

Default entry point

apple-text-audit Natural-language audit entry point for most review requests.

Skills Browse the rest of the catalog when the work is not a code scan.

If the user just asked for a review in natural language, start with apple-text-audit. Reach for the agent page when you want a direct tool handoff instead.