Supercharge Tech Writing with Claude Code Subagents and Agent Skills
Author: Kaz Sato — Staff Developer Advocate, Google Cloud AI Publication: Google Cloud — Community (Medium) Date: October 2025
How to maintain high-quality technical documentation that’s technically accurate, well-written, and consistent — using Claude Code subagents and agent skills.
The Problem
As a developer advocate, maintaining quality technical docs is a challenge: technically accurate, well-written, consistent with the latest updates, and covering multiple aspects (code correctness, API usage, formatting, clarity).
The Solution
Claude Code’s subagents + agent skills for automated documentation review.
What Are Subagents?
Specialized agents defined in .claude/agents/ — each focused on a specific review aspect (docs style, code correctness, API compliance).
What Are Agent Skills?
Agent Skills provide subagents with access to specific knowledge bases — documentation, source code, or API references. This gives them deep, contextual understanding of your technology stack.
The Strategy
Kaz defined two specialized subagents:
1. docs-reviewer Agent
- Focus: Documentation formatting, structure, writing quality
- Example 1: Fixing incomplete imports in code samples
- Example 2: Fixing inconsistent heading levels
2. adk-reviewer Agent
- Focus: Code correctness, API usage, ADK-specific patterns
- Example 3: Fixing deprecated API usage
- Example 4: Deep dive into streaming behavior
Key Takeaways
- Define one subagent per review dimension (style, correctness, consistency)
- Give each agent the right skills (docs, source code, API refs)
- Run all agents against the same doc for comprehensive coverage
- Subagents catch issues human reviewers miss (deprecated APIs, inconsistent patterns across docs)
Getting Started
# Define subagents in .claude/agents/
# Define skills in .claude/skills/
# Run the review pipeline