skills.rest
jcdendrite
jcdendriteCOMMUNITY
Explore all skills by this author →
GitHub

agent-review

Audit Claude Code agent files for contract and trigger correctness.

4|1|Updated Mar 11, 2026
Search Tags:#claude code#frontmatter#quality checklist#agent audit#trigger validation#deterministic behavior#configuration governance
One-click install
npx skills add https://github.com/jcdendrite/claude-config --skill agent-review-jcdendrite
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-review
Source: https://github.com/jcdendrite/claude-config/tree/main/claude/.claude/skills/agent-review
Command: npx skills add https://github.com/jcdendrite/claude-config --skill agent-review-jcdendrite

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

It helps you audit Claude Code agent files for contract violations, risky trigger design, and structural/behavior issues that can cause agents to misfire or behave inconsistently.

Core Features & Use Cases

  • Frontmatter & dispatch contract validation: checks that required fields (like name and description) exist and that agent options are correctly shaped for the harness.
  • Trigger accuracy and non-trigger discipline: ensures descriptions include precise TRIGGER when: and DO NOT TRIGGER when: blocks to avoid accidental delegation.
  • Operational instruction quality: verifies voice, structure, and checklist formatting so the agent’s behavior is deterministic and reviewable.
  • Length and behavior tests: flags narrative bloat and validates that each line carries behavior-changing instructions.
  • Cross-reference and duplication governance: enforces correct pointers to adjacent skills/agents and requires justification for duplicated content.

Quick Start

Ask the agent-review skill to audit the changes in your updated .claude/agents/*.md (or plugins/*/agents/*.md) and report any contract, trigger, or behavior-test failures it finds.

Frequently Asked Questions about agent-review

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

FAQPage Schema
How do I audit Claude Code agent files for trigger accuracy and frontmatter errors?▼

Auditing Claude Code agent files involves validating YAML frontmatter contracts, enforcing strict TRIGGER when: and DO NOT TRIGGER when: blocks, and checking operational voice to prevent accidental delegation and ensure deterministic behavior.

What causes Claude Code agents to misfire or behave inconsistently during dispatch?▼

Agents misfire due to contract violations, risky trigger design, and structural issues like missing description fields, imprecise trigger blocks, narrative bloat, or duplicated content without justification across the roster.

How do I validate the YAML frontmatter contract for my .claude/agents/*.md files?▼

Validate frontmatter by checking that required fields like name and description exist, ensuring agent options are correctly shaped for the harness, and enforcing per-line behavior tests alongside cross-reference consistency.

Does this agent review process check for duplicated content across a specialist roster?▼

Yes, the agent audit enforces cross-reference and duplication governance by checking correct pointers to adjacent skills or agents and requiring explicit justification for any duplicated content found across specialist files.

How do I ensure deterministic behavior in Claude Code agent configuration files?▼

Ensure deterministic behavior by verifying operational instruction quality, validating that each line carries behavior-changing instructions, and flagging narrative bloat that reduces the reviewability of agent files.

Can I audit agent files located inside plugins directories as well as .claude/agents?▼

Yes, the audit applies to reviewing agent roster entries in both .claude/agents/*.md and plugins/*/agents/*.md, verifying trigger accuracy and structural consistency across the entire specialist set.

Related Skills

Explore More Software Engineering Skills→
jpoutrinjpoutrin
14

create-agent

Create Claude Code agent markdown files with required YAML frontmatter fields.

Community
Basic
sigitpoerwosigitpoerwo

behavioral-modes

Switches AI operational modes to align behavior with the task.

Community
Intermediate
jh941213jh941213
125

harness-audit

Diagnose Claude Code harness health across eight evaluation dimensions.

Community
Advanced
gupsammygupsammy
270

create-agent

Generate Claude Code agent files with YAML frontmatter and system prompts.

Community
Intermediate