configs-variations

Experiment with configs by creating and managing variations. Helps you test different models, prompts, and parameters to find what works best through systematic experimentation.

2.75x

Quality

55%

Does it follow best practices?

Impact

99%

2.75x

Average score across 3 eval scenarios

Securityby

Passed

No known issues

Fix and improve this skill with Tessl

tessl review fix ./skills/agentcontrol/configs-variations/SKILL.md

Quality

Content

77%

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

This is a solid, actionable skill that clearly guides Claude through config variation experimentation using specific MCP tools. Its main strengths are the precise tool usage instructions, the clone-with-overrides workflow, and the strong safety guardrails around baseline protection. Weaknesses include some redundancy between sections (Safety vs. What NOT to Do vs. Step 3 guidance) and mild verbosity in explaining principles Claude already understands.

Suggestions

Consolidate the 'Safety: Protect the Baseline' and 'What NOT to Do' sections to eliminate redundancy — the baseline protection rules appear in both places.

Remove or significantly trim the 'Core Principles' section, as 'test one thing at a time' and 'have a hypothesis' are general experimental design concepts Claude already knows — the skill already enforces these through its concrete workflow instructions.

Dimension	Reasoning	Score
Conciseness	The skill is mostly efficient but includes some unnecessary content. The 'Core Principles' section states things Claude already knows (test one thing at a time, have a hypothesis). The 'What NOT to Do' section partially duplicates the Safety section and Step 3 guidance. The workflow table in Step 2 is helpful but the surrounding prose could be tighter.	2 / 3
Actionability	The skill provides concrete, specific guidance on which MCP tools to use, exact field names to pass, the modelConfigKey format with real examples, and clear instructions on what to override vs. leave unset. While there's no executable code per se, this is an instruction-only skill for MCP tool usage and the guidance is highly specific and actionable.	3 / 3
Workflow Clarity	The workflow is clearly sequenced (Steps 1-4) with explicit verification in Step 4, a clear alternative path in Step 3, and strong safety guardrails about protecting the baseline. The clone tool's built-in diff (returning source and created variation) serves as a validation checkpoint. The 'Note on API responses' provides error-recovery guidance.	3 / 3
Progressive Disclosure	The content is well-structured with clear sections and headers, and references related skills at the bottom. However, the document is fairly long (~100 lines of substantive content) and some sections like the detailed Safety rules and What NOT to Do list could potentially be split out. With no bundle files, there's no external reference structure to leverage, but the content is borderline monolithic for a single file.	2 / 3
	Total	10 / 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 provides a moderate sense of what the skill does—managing configuration variations for experimentation—but lacks explicit trigger guidance ('Use when...') and concrete specific actions. It uses second person voice ('Helps you') which violates the third-person requirement, and the terminology is broad enough to potentially conflict with related skills around prompt engineering or model configuration.

Suggestions

Add an explicit 'Use when...' clause with trigger terms like 'experiment with configs', 'compare model settings', 'test prompt variations', 'config sweep', or 'parameter tuning'.

Rewrite in third person voice (e.g., 'Creates and manages configuration variations' instead of 'Helps you test').

List more specific concrete actions such as 'generate config variants, run comparisons across models, track experiment results, and identify optimal parameter combinations'.

Dimension	Reasoning	Score
Specificity	Names the domain (configs/experimentation) and some actions ('creating and managing variations', 'test different models, prompts, and parameters'), but lacks concrete specific actions like 'create A/B test configs', 'compare results', or 'generate variation matrices'.	2 / 3
Completeness	Describes what it does (creating/managing config variations for experimentation) but has no explicit 'Use when...' clause or trigger guidance. Per rubric guidelines, a missing 'Use when...' clause caps completeness at 2, and the 'when' is only vaguely implied, making this a weak 1-2. Additionally, the description uses second person ('Helps you'), which is penalized.	1 / 3
Trigger Term Quality	Includes some relevant keywords like 'configs', 'models', 'prompts', 'parameters', 'experimentation', and 'variations', but misses common user terms like 'A/B test', 'hyperparameter tuning', 'config sweep', 'compare configurations', or 'benchmark'.	2 / 3
Distinctiveness Conflict Risk	The concept of 'config variations' and 'systematic experimentation' provides some specificity, but terms like 'models', 'prompts', and 'parameters' are broad enough to overlap with general LLM configuration, prompt engineering, or model evaluation skills.	2 / 3
	Total	7 / 12 Passed

Validation

100%

Warnings & errors only

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

Validation — 11 / 11 Passed

Validation for skill structure

No warnings or errors.

Repository: launchdarkly/ai-tooling
Commit: 913b745

Reviewed: 19 days ago

Table of Contents

Discovery Implementation Validation

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.