Agent Skills › different-ai/openwork › run-evals

run-evals

GitHub

兼容旧的自动化评估运行器，仅在用户明确请求运行已存在的evals/flows目录下的流文件时使用。禁止修改或创建新流，仅负责列出和运行现有流并报告失败情况。

.opencode/skills/run-evals/SKILL.md different-ai/openwork

Trigger Scenarios

用户指定运行一个已存在的 evals/flows/*.flow 文件用户明确要求使用 legacy eval runner

Install

npx skills add different-ai/openwork --skill run-evals -g -y

More Options

Non-standard path

npx skills add https://github.com/different-ai/openwork/tree/dev/.opencode/skills/run-evals -g -y

Use without installing

npx skills use different-ai/openwork@run-evals

指定 Agent (Claude Code)

npx skills add different-ai/openwork --skill run-evals -a claude-code -g -y

安装 repo 全部 skill

npx skills add different-ai/openwork --all -g -y

预览 repo 内 skill

npx skills add different-ai/openwork --list

SKILL.md

Frontmatter

{
    "name": "run-evals",
    "description": "DEPRECATED legacy automation runner. Load only when a user explicitly asks to run an EXISTING evals\/flows\/*.flow file. New coverage uses run-tests."
}

Legacy Eval Runner Compatibility

Use this guide only when the user names a flow that already exists in evals/flows/. The directory is frozen.

For current verification, use prove-a-pr → write-a-spec → run-tests → publish-evidence.

Refuse to create, scaffold, copy, rename, or modify a legacy flow.
If the requested behavior has no existing flow, use write-a-spec and run-tests; new specs import test from @openwork/testkit.
Manual browser work is debugging, not replacement verdict evidence.

List and run only the requested existing flow:

pnpm evals:legacy --list
pnpm evals:legacy --flow <existing-id> --cdp-url <electron-cdp-url>

Use daytona-electron-test only when the existing flow needs a Daytona app. Report failures and obsolete coverage without changing the flow. See the fraimz compatibility skill only when the user explicitly requests the legacy demo-mode artifact.