Self Aware Prediction System

Provides predictions with quantified uncertainty by evaluating data completeness, prediction type, and confidence to inform decision-making risks.

Security Scan

The skill's files, instructions, and small local script are consistent with its stated purpose of quantifying prediction uncertainty and do not request credentials, external installs, or network access.

Purpose & Capability

Name/description (quantifying prediction uncertainty) matches the included docs and the single assessment script. There are no unrelated requirements (no credentials, binaries, or config paths).

Instruction Scope

SKILL.md stays on-topic (assess information completeness, prediction type, run the uncertainty assessment). It does not instruct reading unrelated files, exfiltrating data, or contacting external endpoints.

Install Mechanism

No install spec (instruction-only skill aside from a small included script). Nothing is downloaded or written to disk at install time by the skill itself.

Credentials

The skill declares no required environment variables, credentials, or config paths. The included Python script is local and does not access environment variables or external services.

Persistence & Privilege

always is false, the skill does not request persistent/system-wide changes or elevated privileges, and it does not modify other skills' configurations.

Assessment

This skill appears internally coherent and low-risk: it contains documentation and a small local Python function that computes a simple confidence score based on declared inputs. Before installing or enabling for automated use, consider: 1) test the script on representative inputs to confirm its behavior and that its simple heuristics match your needs (it uses a fixed base score and static weights); 2) verify there are no hidden files or later updates that introduce network calls or credential usage; 3) if you plan to supply sensitive content to the skill, ensure your agent's privacy settings are appropriate — the skill itself does not exfiltrate data but your agent may log or route inputs elsewhere; 4) if you need stronger uncertainty quantification, review or replace the heuristic with a more rigorous/statistical method. Overall, nothing in the package requests disproportionate access or contradicts its described purpose.

自知之明预测系统 (Self-Aware Prediction System)

核心理念

• 不追求预测完美度，而是量化预测的不确定性
• 承认自己的知识边界和认知局限
• 为每个预测输出置信区间和可信度评分

预测分类框架

1. 信息完整性评估

• 高完整性: 有足够的数据和事实支撑
• 中完整性: 部分信息缺失，需要推理
• 低完整性: 严重信息不足，纯推测

2. 预测类型分类

• 事实型: 基于现有数据的客观预测
• 趋势型: 基于历史模式的趋势推断
• 概率型: 多种可能性的概率分布
• 推测型: 高度不确定的主观推测

不确定性量化机制

置信度评分 (0-10分)

• 1-3分: 严重不确定，建议查询更多信息
• 4-6分: 部分不确定，需要谨慎对待
• 7-8分: 相对确定，但仍需注意边界
• 9-10分: 高度确定，基于充分证据

风险警示触发条件

• 预测类型为"推测型"
• 信息完整性为"低"
• 置信度低于5分
• 涉及重大决策影响

应用场景

• 回答用户问题时评估自己的确定性
• 预测未来任务完成情况
• 评估建议的可行性
• 判断是否需要寻求外部信息