Loading...
Loading...
Found 45 Skills
Benchmark any agent skill to measure whether it actually improves performance. Use when the user wants to evaluate, test, or compare a skill against baseline, or when they mention "benchmark", "eval", "skill performance", or "does this skill help". Runs isolated eval sessions with and without the skill, grades outputs via layered grading (deterministic checks + LLM-as-judge), analyzes behavioral signals, and generates a comparison report with a USE / DON'T USE verdict.
This skill should be used when the user wants to run baseline evaluations on existing agent skills, regenerate transcripts after a model upgrade, or check whether a skill still solves the gap it was authored for. Common triggers include "rerun the baselines", "re-eval skill X", "test all the skills", "check for skill drift", and "run the evals". Bakes in verbatim transcript capture (no paraphrasing), deterministic-only grading (regex / contains / file_exists — no LLM-as-judge), and the iteration-N workspace convention. Skip when authoring a new skill (use skill-creator) or modifying skill content directly.
Implement comprehensive evaluation strategies for LLM applications using automated metrics, human feedback, and benchmarking. Use when testing LLM performance, measuring AI application quality, or establishing evaluation frameworks.
Launch a meta-judge then a judge sub-agent to evaluate results produced in the current conversation
Build evaluation frameworks for agent systems. Use when testing agent performance, validating context engineering choices, or measuring improvements over time.
Attach judges to AI Config variations for automatic LLM-as-a-judge evaluation. Create custom judges, configure sampling rates, and monitor quality scores.
Evaluate and improve Claude Code commands, skills, and agents. Use when testing prompt effectiveness, validating context engineering choices, or measuring improvement quality.
Execute a task with sub-agent implementation and LLM-as-a-judge verification with automatic retry loop
Create validated LLM-as-a-Judge evaluators following best practices — binary Pass/Fail judges with TPR/TNR validation for measuring specific failure modes. Use when you need to automate quality checks, build guardrails, or measure a specific failure mode identified during trace analysis. Do NOT use when failures are fixable with prompt changes (use optimize-prompt) or when failure modes are unknown (use analyze-trace-failures first).
Iteratively inspect traces, interview the user, and create LangSmith online evaluators one at a time. Use specifically for creating online evaluators for use within LangSmith -- use "eval-engineering" for Harbor-style online evaluations.
Generates a Jupyter notebook that evaluates a fine-tuned SageMaker model using LLM-as-a-Judge. Use when the user says "evaluate my model", "how did my model perform", "compare models", or after a training job completes. Supports built-in and custom evaluation metrics, evaluation dataset setup, and judge model selection.
Use when evaluating agent performance, building test frameworks, measuring quality, or asking about "agent evaluation", "LLM-as-judge", "agent testing", "quality metrics", "evaluation rubrics", "agent benchmarks"