xml-standards
The xml-standards Claude Code skill provides structured XML tag templates and patterns for building Claude Code agents and orchestrator commands. Use this skill when designing new agents or implementing command structures to maintain consistency with Anthropic best practices, ensure proper role definition, constraint handling, workflow phases, knowledge organization, examples, and task delegation across agent types.
git clone --depth 1 https://github.com/MadAppGang/claude-code /tmp/xml-standards && cp -r /tmp/xml-standards/plugins/agentdev/skills/xml-standards ~/.claude/skills/xml-standardsSKILL.md
plugin: agentdev
updated: 2026-01-20
# XML Tag Standards
## Core Tags (Required for ALL Agents/Commands)
### `<role>`
Defines agent identity and purpose.
```xml
<role>
<identity>Expert [Domain] Specialist</identity>
<expertise>
- Core skill 1
- Core skill 2
- Core skill 3
</expertise>
<mission>
Clear statement of what this agent accomplishes
</mission>
</role>
```
### `<instructions>`
Defines behavior constraints and workflow.
```xml
<instructions>
<critical_constraints>
<constraint_name>
Description of critical rule that must be followed
</constraint_name>
<todowrite_requirement>
You MUST use Tasks to track workflow progress.
</todowrite_requirement>
</critical_constraints>
<core_principles>
<principle name="Name" priority="critical|high|medium">
Description of principle
</principle>
</core_principles>
<workflow>
<phase number="1" name="Phase Name">
<step>Step description</step>
<step>Step description</step>
</phase>
</workflow>
</instructions>
```
### `<knowledge>`
Domain-specific best practices and templates.
```xml
<knowledge>
<section_name>
Best practices, patterns, or reference material
</section_name>
<templates>
<template name="Template Name">
Template content
</template>
</templates>
</knowledge>
```
### `<examples>`
Concrete usage scenarios (2-4 required).
```xml
<examples>
<example name="Descriptive Name">
<user_request>What user asks for</user_request>
<correct_approach>
1. Step one
2. Step two
3. Step three
</correct_approach>
</example>
</examples>
```
### `<formatting>`
Communication style and output format.
```xml
<formatting>
<communication_style>
- Style guideline 1
- Style guideline 2
</communication_style>
<completion_message_template>
Template for completion messages
</completion_message_template>
</formatting>
```
---
## Specialized Tags by Agent Type
### Orchestrators (Commands)
```xml
<orchestration>
<allowed_tools>Task, Bash, Read, TaskCreate, TaskUpdate, TaskList, TaskGet, AskUserQuestion</allowed_tools>
<forbidden_tools>Write, Edit</forbidden_tools>
<delegation_rules>
<rule scope="design">ALL design → architect agent</rule>
<rule scope="implementation">ALL implementation → developer agent</rule>
<rule scope="review">ALL reviews → reviewer agent</rule>
</delegation_rules>
<phases>
<phase number="1" name="Phase Name">
<objective>What this phase achieves</objective>
<steps>
<step>Step description</step>
</steps>
<quality_gate>Exit criteria for this phase</quality_gate>
</phase>
</phases>
</orchestration>
<error_recovery>
<strategy>
Recovery steps for common failures
</strategy>
</error_recovery>
```
### Planners (Architects)
```xml
<planning_methodology>
<approach>How planning is performed</approach>
<deliverables>What planning produces</deliverables>
</planning_methodology>
<gap_analysis>
<checklist>Items to verify during planning</checklist>
</gap_analysis>
<output_structure>
<format>Structure of planning output</format>
</output_structure>
```
### Implementers (Developers)
```xml
<implementation_standards>
<file_writing_standards>
<standard name="Standard Name">Description</standard>
</file_writing_standards>
<quality_checks mandatory="true">
<check name="check_name" order="1">
<tool>Tool name</tool>
<command>Command to run</command>
<requirement>What must pass</requirement>
<on_failure>Recovery action</on_failure>
</check>
</quality_checks>
<validation_checks>
<check order="1" name="Check Name">
Validation criteria
</check>
</validation_checks>
</implementation_standards>
```
### Reviewers
```xml
<review_criteria>
<focus_areas>
<area name="Area Name" priority="critical|high|medium" weight="20%">
**Check:**
- Item to verify
- Item to verify
**Common Issues:**
- Issue description
**Critical if**: Condition for critical severity
**High if**: Condition for high severity
</area>
</focus_areas>
<feedback_format>
Template for review feedback
</feedback_format>
</review_criteria>
<approval_criteria>
<status name="PASS">Criteria for passing</status>
<status name="CONDITIONAL">Criteria for conditional approval</status>
<status name="FAIL">Criteria for failure</status>
</approval_criteria>
```
### Testers
```xml
<testing_strategy>
<approach>Testing methodology</approach>
<test_types>
<type name="Type Name">Description</type>
</test_types>
</testing_strategy>
<coverage_requirements>
<requirement>Coverage criteria</requirement>
</coverage_requirements>
```
---
## Nesting Rules
1. **Proper Hierarchy** - Tags must be properly nested
2. **Closing Tags** - All opening tags must have closing tags
3. **Semantic Attributes** - Use `name`, `priority`, `order` attributes
4. **Consistent Naming** - Use lowercase-with-hyphens for tag names
## Code Blocks in XML
```xml
<template name="Example">
```language
// code here - note: opening ``` directly under tag
```
</template>
```
## Character Escaping
Only in XML attribute values and text nodes (NOT in code blocks):
- `<` for `<`
- `>` for `>`
- `&` for `&`|
|
|
Common agent patterns and templates for Claude Code. Use when implementing agents to follow proven patterns for Tasks integration, quality checks, and external model invocation via claudish CLI.
YAML frontmatter schemas for Claude Code agents and commands. Use when creating or validating agent/command files.
YAML format for Claude Code agent definitions as alternative to markdown. Use when creating agents with YAML, converting markdown agents to YAML, or validating YAML agent schemas. Trigger keywords - "YAML agent", "agent YAML", "YAML format", "agent schema", "YAML definition", "convert to YAML".
Linear API patterns and examples for autopilot. Includes authentication, webhooks, issue CRUD, state transitions, file attachments, and comment handling.
Proof artifact generation patterns for task validation. Covers screenshots, test results, deployments, and confidence scoring.