← 返回 Skills 市场
Pa Eval
作者
Netanel Abergel
· GitHub ↗
· v1.0.1
· MIT-0
136
总下载
0
收藏
1
当前安装
2
版本数
在 OpenClaw 中安装
/install pa-eval
功能描述
Evaluate PA performance through structured scoring, owner feedback analysis, and behavioral benchmarking. Use when: conducting a weekly/monthly PA performanc...
安全使用建议
This skill appears coherent and minimal: it only creates local eval files and scores PA performance. Before installing, confirm two operational details: (1) what channels/events the agent will monitor to detect owner feedback (so you avoid unintended message monitoring), and (2) file location/retention and permissions for $HOME/.openclaw/workspace/.learnings/eval. If you want stricter control, disable automatic runs and require manual invocation for weekly/monthly reports.
功能分析
Type: OpenClaw Skill
Name: pa-eval
Version: 1.0.1
The skill bundle is a performance evaluation tool for an AI agent, designed to track metrics like accuracy and speed. It includes a simple bash script in SKILL.md that creates a local directory (~/.openclaw/workspace/.learnings/eval) and writes markdown templates for logging, which is consistent with its stated purpose and lacks any indicators of malicious intent or data exfiltration.
能力评估
Purpose & Capability
The name/description (PA performance scoring, feedback, benchmarks) match the instructions: templates, weekly/monthly aggregation, and simple benchmark calculations. No unrelated binaries, cloud credentials, or surprising permissions are requested.
Instruction Scope
Instructions tell the agent to create weekly eval files under $HOME/.openclaw/workspace/.learnings/eval, score dimensions, run monthly benchmarks, and 'log feedback signals automatically when detected.' The file writes and scoring are appropriate. The only ambiguity: 'log automatically when detected' and reaction/text signals assume the agent can monitor owner messages/reactions — the SKILL.md doesn't limit channels or explain how signals are detected. Confirm monitoring scope and privacy before enabling automatic logging.
Install Mechanism
Instruction-only skill with no install spec, no downloads, and no code files to execute beyond the provided small shell snippet; lowest-risk installation footprint.
Credentials
No environment variables, credentials, or config paths are requested. The requested filesystem write is a single per-user workspace path under $HOME, which is proportionate to the stated purpose.
Persistence & Privilege
Metadata sets always:false (not forced into all agents). The doc repeatedly says 'Run automatically' (every 7 days, 'log immediately'), which implies autonomous scheduling or event monitoring — this is reasonable for the skill's purpose but requires you to accept that the agent will autonomously create logs and monitor owner interactions if enabled. Consider limiting triggers or requiring explicit owner confirmation before automatic runs.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install pa-eval - 安装完成后，直接呼叫该 Skill 的名称或使用
/pa-eval触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.1
reactions rule, close-the-loop, reply-to rules; skill-master analytics hook; skill-analytics added
v1.0.0
- Initial release of the pa-eval skill for structured PA performance evaluation.
- Provides weekly, correction-based, monthly, and on-demand performance reviews using a standardized scoring system (Execution, Accuracy, Speed, Proactivity, Communication, Memory, Tool Use, Judgment).
- Includes markdown templates for weekly self-evaluation and monthly reports, with grading and actionable feedback sections.
- Automates logging and analysis of owner feedback signals (positive/negative reactions, corrections, initiative/memory gaps).
- Benchmarks key performance metrics (task completion, accuracy, memory retention) and provides cost-saving usage tips.
元数据
常见问题
Pa Eval 是什么？
Evaluate PA performance through structured scoring, owner feedback analysis, and behavioral benchmarking. Use when: conducting a weekly/monthly PA performanc... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 136 次。
如何安装 Pa Eval？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install pa-eval」即可一键安装，无需额外配置。
Pa Eval 是免费的吗？
是的，Pa Eval 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Pa Eval 支持哪些平台？
Pa Eval 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Pa Eval？
由 Netanel Abergel（@netanel-abergel）开发并维护，当前版本 v1.0.1。
推荐 Skills