← 返回 Skills 市场
866
总下载
2
收藏
6
当前安装
1
版本数
在 OpenClaw 中安装
/install compare
功能描述
Rigorous comparisons with confidence parity, weighted criteria, and research depth tracking.
使用说明 (SKILL.md)
Core Principle
Comparisons fail when confidence is uneven. Only as reliable as the weakest-researched dimension.
Protocol
Criteria → Research Parity → Confidence Check → Score → Present
1. Criteria
- Load domain defaults (
domains.md) - Overlay user preferences from memory
- If unknown: "What matters most here?"
- Output: Ranked criteria with weights (sum = 100%)
2. Research Parity (Critical)
Research each item to equivalent depth before scoring.
Track: | Criterion | Item A sources | Item B sources |
5 reviews for A but 1 for B? Research more for B first. Never score unbalanced data.
3. Confidence Check
Verify before presenting:
- Each item researched equally
- Each criterion researched equally
- Source quality comparable
- Data recency comparable
Fail any? Research more OR caveat explicitly.
4. Score
Final = Σ(criterion_score × weight) — Show the math.
5. Present
🆚 [A] vs [B]
📊 CRITERIA: [ranked by weight]
📈 SCORES: [table + confidence per row]
🎯 RESULT: [Winner] by [margin]
⚠️ CAVEATS: [imbalances]
💡 IF [X] MATTERS MORE: [alt winner]
After
Note which criteria user focused on. Update preferences.md by category.
Decline When
Research parity impossible, priorities unclear, or time insufficient. Partial > misleading.
References: domains.md, confidence.md, traps.md, preferences.md
安全使用建议
This skill appears internally consistent and safe to install, but consider these practical points before using it: (1) it will perform external research (web/API calls) to collect sources — avoid giving it items that require confidential or proprietary research unless you trust the agent's browsing/data-access settings; (2) it reads agent memory to overlay user preferences, so review what is stored in memory if you care about privacy; (3) it updates preferences.md to persist learned priorities — if you don’t want persistent preference changes, back up or inspect that file after runs; (4) pay attention to the sources it cites (quality and recency) — the skill emphasizes parity, but quality of sources still matters; (5) if you prefer to approve each run, restrict autonomous invocation or call the skill only when needed. Overall: coherent and proportional to its purpose.
功能分析
Type: OpenClaw Skill
Name: compare
Version: 1.0.0
The OpenClaw AgentSkills bundle 'compare' consists entirely of documentation and configuration files (`_meta.json`, `SKILL.md`, `confidence.md`, `domains.md`, `preferences.md`, `traps.md`). The `SKILL.md` file provides instructions for the AI agent on how to perform rigorous comparisons, including loading default criteria from `domains.md` and updating user preferences in `preferences.md`. All file operations are local and serve the stated purpose of the skill. There is no evidence of data exfiltration, malicious execution, persistence mechanisms, or prompt injection designed to subvert the agent's core function or harm the user. The content is well-aligned with its described purpose.
能力评估
Purpose & Capability
Name/description (rigorous comparisons) aligns with the included files and runtime instructions. The skill only references local domain defaults, confidence rules, traps, and preference tracking — all consistent with producing balanced comparisons.
Instruction Scope
Instructions are explicit and scoped to: (a) load domains.md, (b) consult user preferences from memory, (c) research both items to equal depth, (d) compute weighted scores, and (e) update preferences.md. This is coherent. Note: the skill expects the agent to perform external research (web/API calls) and to read/merge agent memory; both are necessary for the stated goal but mean the agent will fetch external sources and access stored preferences. It also writes to preferences.md to persist learned priorities.
Install Mechanism
Instruction-only skill with no install spec, no binaries, and no code files. Lowest-risk install profile.
Credentials
Requires no environment variables, no credentials, and no system config paths. The declared requirements are minimal and appropriate for the function.
Persistence & Privilege
always:false (normal). The skill writes/updates its local preferences.md and reads agent memory — reasonable for learning user preferences but worth noting because it persists learned settings and reads stored memory data.
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 compare - 安装完成后，直接呼叫该 Skill 的名称或使用
/compare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
v1.0.0
Initial release
元数据
常见问题
Compare 是什么？
Rigorous comparisons with confidence parity, weighted criteria, and research depth tracking. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 866 次。
如何安装 Compare？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install compare」即可一键安装，无需额外配置。
Compare 是免费的吗？
是的，Compare 完全免费（开源免费），可自由下载、安装和使用。
Compare 支持哪些平台？
Compare 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 Compare？
由 Iván（@ivangdavila）开发并维护，当前版本 v1.0.0。
推荐 Skills