Token导航 LogoToken导航TokenDH.com
研究检索只读github未标认证来源可访问许可证需确认审计通过

self-reflect自我反思

Agent Skill

self-reflect 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

380

周安装

16

GitHub Stars

40

下载量

133
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:self-reflect(自我反思)
来源仓库:https://github.com/knoopx/pi
仓库路径:skills/self-reflect
安装命令:
npx skills add https://github.com/knoopx/pi --skill self-reflect
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/knoopx/pi --skill self-reflect

简介

self-reflect 用于查找、检索和筛选相关信息,支持快速定位候选结果。

  • 适用于基于关键词、任务场景或来源线索的信息筛选需求。
  • 通过 npx skills add 命令安装,建议结合原始 README 了解具体实现。
  • 安装前应确认权限范围和维护状态,避免触发不必要的网络或文件操作。
  • 涉及命令执行时需注意运行目录和数据边界。

SKILL.md

Self-Reflect

Reflect on the current session to extract learnings and improvements.

Goals

GoalMethod
Failure AnalysisReview tool errors and retries
Efficiency ReviewFind redundant or wasteful steps
Pattern ExtractionDiscover reusable workflows
Knowledge GapsNote missing skills or context

What to Look For

Failures

  • Tool calls that errored
  • Multiple retries of same action
  • User corrections or clarifications
  • Misunderstood requirements

Inefficiencies

  • Read file multiple times unnecessarily
  • Overly large edits that could be smaller
  • Commands that could be combined
  • Backtracking on approach

Patterns

  • Multi-step workflows worth automating
  • Repeated prompt structures
  • Common tool sequences

Categorize Findings

CategoryAction
Tool MisuseUpdate AGENTS.md with pattern
Missing SkillCreate new skill
Repeated StepsCreate prompt template
Edge CaseAdd to skill or guardrail

Output Files

FilePurpose
agent/skills/<name>/Multi-step workflows
agent/prompts/<name>.mdReusable prompt templates
AGENTS.mdProject-specific rules
agent/APPEND_SYSTEM.mdUniversal agent manifesto

APPEND_SYSTEM.md Format

New sections use numbered Roman headings with Against or For stance (## N. Against X or ## N. For X). Prose paragraphs, no bullet lists, no bold markers, no indented context blocks. Declarative statements in first person. General principles, no project-specific tool names or examples. Read existing sections before adding new ones — match the style exactly.

Common Issues

Edit Failures

PatternLearning
"Could not find"Always read file before editing
Multiple edit retriesUse smaller, targeted edits
Wrong whitespaceMatch exact indentation/quotes

Bash Failures

PatternLearning
grep exit 1Use `grep... \\true`
Command not foundCheck available tools first
Permission deniedNote required permissions

Workflow Issues

PatternLearning
Asked same questionClarify requirements upfront
Backtracked on designPlan before implementing
Missed edge caseAdd to skill checklist

Reflection Output

## Session Summary

**Task:** [What was accomplished]

**Failures:**

- [Tool/approach that failed and why]

**Inefficiencies:**

- [Steps that could be optimized]

**Learnings:**

- [New patterns or knowledge gained]

**Improvements:**

- [ ] [Concrete change to make]

Related Skills

  • self-improve: Analyze historical sessions for patterns
  • skill-authoring: Write discovered skills

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

38%
按下载量换算51

Claude

27.57%
按下载量换算37

Cursor

18.35%
按下载量换算24

Gemini CLI

9.13%
按下载量换算12

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills