google-agents-cli-eval

The google-agents-cli-eval skill provides a command-line interface for evaluating AI agents, including dataset synthesis, trace generation, grading, failure analysis, and prompt optimization.

47.2K
Installs
6
Use cases
5/10
Quality

Is google-agents-cli-eval safe to install?

Review the source first

Review the source first: our audit of google-agents-cli-eval's source files found 10 shell commands, 0 external URLs, file reads and writes (high risk). Every command and URL listed appears verbatim in the skill's source. The skill executes shell commands via the agents-cli binary and performs file system operations to read datasets and write evaluation traces and results.

How we audit skills: our security review methodology.

Who is this skill for?

Developers building and iterating on AI agents using the Google Agent Platform.

What can you do with it?

  • Synthesize evaluation datasets for multi-turn conversations
  • Generate agent traces over evaluation datasets
  • Grade agent traces using built-in or custom metrics
  • Analyze agent failure modes through LLM-based clustering
  • Optimize agent prompts using ADK GEPA
  • Compare evaluation results to track performance improvements

How good is this skill?

Quality score: 5/10. The documentation is comprehensive, providing clear workflows, command examples, and guidance on the evaluation lifecycle.

What does the skill file contain?

SKILL.md
# Agent Evaluation Guide

> **Requires:** `agents-cli` (`uv tool install google-agents-cli`) — [install uv](https://docs.astral.sh/uv/getting-started/installation/index.md) first if needed.

> **Scaffolded project?** If you used `/google-agents-cli-scaffold`, you already have `agents-cli eval run` (chains `generate` + `grade`), `tests/eval/datasets/`, and `tests/eval/eval_config.yaml`. Start with executing `eval run` and iterate from there.

## Reference Files

| File | Contents |
|------|----------|
| `references/dataset_schema.md` | Canonical EvaluationDataset schema — all field types, JSON ...

Frequently asked questions

How do I start evaluating an agent?

If using a scaffolded project, execute agents-cli eval run to chain generation and grading.

What should I do if my agent fails an evaluation?

Inspect the generated JSON or HTML results for metric scores and judge rationales, then adjust prompts, tool descriptions, or instructions based on the failure type.

Can I use custom metrics?

Yes, you can define custom LLMMetric or CodeExecutionMetric types in your evaluation configuration.

When should I use eval optimize?

Use it only for prompt-only failures after manual iteration, as it is a long-running and expensive process.

Data sourced from google/agents-cli on GitHub. Install counts from skills.sh. The summary and security audit are derived from the skill's source files: every command and URL listed appears verbatim in the source.

Related skills

find-skills

2.3M

Users seeking to extend agent capabilities with specialized tools, workflows, or knowledge packages

The find-skills skill enables agents to search for, discover, and install modular packages from the open agent skills ecosystem using the Skills CLI.

highclipackage-managervercel-labs

agent-browser

506.7K

AI agents and developers requiring programmatic web interaction, exploratory testing, or automation of Electron desktop applications

The agent-browser CLI provides browser automation for AI agents using Chrome or Chromium via CDP. It supports page navigation, form interaction, data extraction, and testing. The tool utilizes accessibility-tree snapshots and element references for interaction.

highbrowser-automationclivercel-labs

microsoft-foundry

429.2K

Developers building, deploying, and managing AI agents on Azure AI Foundry

The Microsoft Foundry skill provides a framework for the end-to-end lifecycle of AI agents, including scaffolding, deployment, evaluation, fine-tuning, and troubleshooting within Azure AI Foundry.

highazureai-agentsmicrosoft

image-to-video

355.0K

Developers and users who need to automate video generation from images using the RunComfy platform

The image-to-video skill routes user requests to specific RunComfy animation models based on intent. It selects HappyHorse 1.0 I2V for general animation, Wan 2.7 for custom-voiceover lip-sync, or Seedance 2.0 Pro for multi-modal composition. The skill executes the RunComfy CLI to process inputs and download generated video files.

highaivideo-generationagentspace-so