OpenSkillIndex← back to search
claude-code

agent-evaluator

skylv-agent-evaluator

install
source · Clone the upstream repo
git clone https://github.com/SKY-lv/agent-evaluator
Claude Code · Install into ~/.claude/skills/
git clone --depth=1 https://github.com/SKY-lv/agent-evaluator ~/.claude/skills/sky-lv-agent-evaluator-agent-evaluator
manifest: SKILL.md
source content

skylv-agent-evaluator

Agent behavior evaluation engine. 5 criteria, scoring, improvement suggestions.

Skill Metadata

  • Slug: skylv-agent-evaluator
  • Version: 1.0.0
  • Description: Evaluate AI agent actions and outputs against 5 criteria: accuracy, efficiency, clarity, safety, helpfulness. Score 0-100 with letter grade and improvement suggestions.
  • Category: agent
  • Trigger Keywords: 
    evaluate
    , 
    assess
    , 
    score
    , 
    agent quality
    , 
    behavior check

Evaluation Criteria (5)

CriterionWeightDescription
Accuracy25%Correctness of information and actions
Efficiency20%Time and resource usage
Clarity15%Clear communication and reasoning
Safety20%No harmful or dangerous actions
Helpfulness20%Value provided to user

Output

  • Score: 0-100
  • Grade: A+ to D
  • Improvements: Suggestions for criteria < 70

Market Data

Top competitor: 

eval
(0.734) — weak competition, high value.

Built by an AI agent that evaluates itself.