CtrlK
BlogDocsLog inGet started
Tessl Logo

create-agent

Comprehensive guide for creating Claude Code agents with proper structure, triggering conditions, system prompts, and validation - combines official Anthropic best practices with proven patterns

41

Quality

43%

Does it follow best practices?

Impact

No eval scenarios have been run

SecuritybySnyk

Passed

No known issues

Fix and improve this skill with Tessl

tessl review fix ./plugins/customaize-agent/skills/create-agent/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

55%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

This skill is highly actionable with excellent workflow clarity, providing complete production-ready examples, concrete validation steps, and a clear multi-step creation process. However, it is severely over-engineered for a SKILL.md file—at 500+ lines it wastes significant context window space with redundant explanations, concepts Claude already knows, and content that should be split into referenced sub-files. The internal contradictions (e.g., description length guidance) and lack of progressive disclosure further weaken its effectiveness.

Suggestions

Reduce content by 60-70%: Remove 'What Are Agents?' section, the concept comparison table, explanations of YAML frontmatter, and other concepts Claude already knows. Keep only the unique conventions and patterns.

Split into multiple files: Move production examples to EXAMPLES.md, validation rules to VALIDATION.md, triggering patterns to PATTERNS.md, and system prompt design to SYSTEM-PROMPT.md. Reference them from a concise SKILL.md overview.

Resolve internal contradictions: The 'Default Agent Standards' section says descriptions should be ONE sentence with no verbose example blocks, but the main content says 200-1000 chars with 2-4 examples. Consolidate into one consistent recommendation.

Remove the 'AI-Assisted Agent Generation' and 'Elite Agent Architect Process' sections—these are meta-instructions about how to prompt an AI, which is redundant when the skill IS the instruction for Claude.

DimensionReasoningScore

Conciseness

Extremely verbose at ~500+ lines. Explains concepts Claude already knows (what agents are, what autonomous subprocesses are, basic YAML frontmatter). Includes extensive tables, redundant examples, and sections like 'What Are Agents?' that waste tokens. The 'Default Agent Standards' section contradicts earlier advice (e.g., says descriptions should be ONE sentence but earlier says 200-1000 chars with examples). Much content could be cut or condensed significantly.

1 / 3

Actionability

Provides fully concrete, copy-paste ready examples including complete agent file structures, bash commands for creating directories and files, complete production examples (code-quality-reviewer, test-generator), specific validation rules with valid/invalid examples, and executable validation scripts. The guidance is specific and actionable throughout.

3 / 3

Workflow Clarity

The 6-step agent creation process (Steps 1-6) is clearly sequenced with explicit validation checkpoints (Step 5: validate with script, Step 6: test triggering). Includes a quality checklist with checkboxes, validation rules table, and specific testing scenarios. The feedback loop of validate → fix → re-validate is implicit but the validation step is explicit and mandatory before deployment.

3 / 3

Progressive Disclosure

Monolithic wall of text with no references to external files despite the content being long enough to warrant splitting. The production examples alone are ~100 lines that could be in a separate EXAMPLES.md. The validation rules, triggering patterns, and system prompt design sections could each be separate referenced files. No bundle files are provided to support this massive single-file approach.

1 / 3

Total

8

/

12

Passed

Description

32%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description identifies its domain (Claude Code agents) and lists some structural elements it covers, but it reads more like a document subtitle than a skill description optimized for selection. It lacks explicit trigger guidance ('Use when...'), uses somewhat generic framing ('comprehensive guide', 'proven patterns'), and misses natural user language variations that would help Claude match it correctly.

Suggestions

Add an explicit 'Use when...' clause with trigger scenarios, e.g., 'Use when the user wants to build a sub-agent, create an agentic workflow, or design multi-agent systems with Claude Code.'

Replace vague phrases like 'comprehensive guide' and 'proven patterns' with specific concrete actions, e.g., 'Defines agent system prompts, configures tool permissions, sets up agent loops, and validates agent behavior.'

Include natural keyword variations users might say: 'sub-agent', 'build an agent', 'agentic', 'multi-agent', 'agent orchestration', 'spawn agent'.

DimensionReasoningScore

Specificity

Names the domain ('Claude Code agents') and some actions ('creating', 'structure', 'triggering conditions', 'system prompts', 'validation'), but these are more like categories than concrete specific actions. It doesn't list discrete operations like 'write system prompts, configure tool permissions, set up agent loops.'

2 / 3

Completeness

Describes what it covers (guide for creating Claude Code agents with structure, prompts, validation) but completely lacks a 'Use when...' clause or any explicit trigger guidance for when Claude should select this skill. Per rubric guidelines, missing 'Use when' should cap completeness at 2, and the 'what' is also only moderately clear, so this scores a 1.

1 / 3

Trigger Term Quality

Includes some relevant terms like 'Claude Code agents', 'system prompts', 'Anthropic best practices', but misses natural user phrases like 'build an agent', 'sub-agent', 'agentic workflow', 'multi-agent', or 'agent orchestration'. The terms present are somewhat technical rather than what users would naturally say.

2 / 3

Distinctiveness Conflict Risk

The focus on 'Claude Code agents' is somewhat distinctive, but 'system prompts' and 'best practices' are broad enough to overlap with general prompt engineering or Claude configuration skills. The phrase 'comprehensive guide' is generic and doesn't help differentiate.

2 / 3

Total

7

/

12

Passed

Validation

72%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation8 / 11 Passed

Validation for skill structure

CriteriaDescriptionResult

skill_md_line_count

SKILL.md is long (670 lines); consider splitting into references/ and linking

Warning

allowed_tools_field

'allowed-tools' contains unusual tool name(s)

Warning

frontmatter_unknown_keys

Unknown frontmatter key(s) found; consider removing or moving to metadata

Warning

Total

8

/

11

Passed

Repository
NeoLabHQ/context-engineering-kit
Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.