Token导航 LogoToken导航TokenDH.com
待分类只读github未标认证来源可访问clear审计异常

close-session闭幕会议

Agent Skill

close-session 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。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

269

周安装

11

GitHub Stars

2

下载量

87
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

3

许可证

MIT

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:close-session(闭幕会议)
来源仓库:https://github.com/abdullahbeam/nexus-design-abdullah
仓库路径:skills/close-session
安装命令:
npx skills add https://github.com/abdullahbeam/nexus-design-abdullah --skill close-session
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。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/abdullahbeam/nexus-design-abdullah --skill close-session

简介

会话终结守护技能,自动保存进度、更新记忆地图并清理临时文件。

  • 适用于防止工作丢失的关键系统级保障机制,非用户主动触发。
  • 读取 steps.md 复选框状态更新项目进展,校验 workspace-map.md 准确性。
  • 生成历史报告前会扫描根目录临时文件,避免残留数据影响下次会话。
  • close-session 属于待分类类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Close Session

Save progress, update memory, regenerate navigation, and ensure system integrity.

Purpose

The close-session skill is the most critical system skill. It ensures nothing is ever lost by:

  • Reading and updating project progress from steps.md checkboxes
  • Validating workspace-map.md accuracy (auto-detect mismatches)
  • Updating memory with decisions and patterns
  • Cleaning temporary files from root folder
  • Creating historical session reports
  • Displaying comprehensive summary

CRITICAL: This skill is AUTO-TRIGGERED by all other skills and projects (not user-dependent!).


Execution Sequence

  1. Initialize TodoWrite with all 10 steps (MANDATORY - prevents skipped steps)
  2. Load workflow.md
  3. Execute steps 1-10 sequentially
  4. Mark each step complete in TodoWrite as you finish it

Critical Rules

  1. TodoWrite is MANDATORY: Initialize at start with all 10 steps - prevents forgetting critical steps
  2. PLANNING phase projects: Skip task completion (Step 2/2.5)
  3. IN_PROGRESS phase projects: Auto-complete if execution signals detected
  4. Session reports: Create in 01-memory/session-reports/
  5. Summary display: ≤5 lines per orchestrator.md rule

Key Features

Automatic & Interactive Task Completion

Smart task completion with automatic detection:

  • Automatic Bulk Complete: If project work completed this session, auto-marks all tasks
  • Manual Bulk Option: If auto-detect missed it, offers bulk-complete during review
  • Interactive Review: Shows first 10 unchecked tasks for manual selection
  • User selects by number ("1, 3, 5"), "all", "bulk complete", or "none"
  • Updates tasks.md automatically (via Edit tool or bulk-complete script)
  • Recalculates progress after any changes

Temp File Cleanup

Interactive cleanup with user choices:

  • Scans root folder for temp files
  • Asks what to do with each: keep, delete, or skip
  • Moves preserved files to project outputs/
  • Reports cleanup summary

Session Reporting

Creates historical record:

  • Generates session report in 01-memory/session-reports/
  • Includes work completed, progress, decisions, patterns
  • Provides context for next session

Progress Tracking

Auto-calculates from checkboxes:

  • Counts total tasks (all - [] and - [x])
  • Counts completed tasks (only - [x])
  • Determines status (PLANNING/IN_PROGRESS/COMPLETE)
  • Identifies next task

Auto-Trigger Support

Called automatically by other skills:

  • create-project
  • validate-system
  • Any skill completion

Memory Preservation

THE critical persistence mechanism:

  • Creates session reports
  • Cleans temp files

Without this skill, context does NOT persist across sessions!


Workflow Overview

Complete workflow with all 9 steps: See workflow.md

Steps (from workflow.md):

  1. Read project state (skip if no IN_PROGRESS projects)
  2. Review task completion (skip if PLANNING phase)
  3. Update maps
  4. Get timestamp
  5. Update memory
  6. Clean temp files (delete.md files not in system folders)
  7. Create session report
  8. Display summary (≤5 lines)
  9. Mark complete
  10. Instruct fresh session

Integration

Auto-Trigger Format

When called by other skills:

Auto-triggering close-session skill...

[Full workflow executes]

Session saved! ✅
[Summary displays]

User-Trigger Format

When user says "done for now":

Closing your session...

[Full workflow executes]

Session saved! ✅
[Summary displays]

All Skills Must End With

Every skill and project workflow should conclude with:

### Final Step: Close Session
Auto-trigger close-session skill to save progress

This ensures:

  • Progress is saved
  • Maps are updated
  • Session is recorded
  • Nothing is lost

Error Handling

For complete error scenarios and solutions, see error-handling.md

Common Scenarios:

No active project → Skip project steps, continue with maps and cleanup

Missing tasks.md → Report in summary, suggest validate-system

Corrupted memory → Rebuild from scan, report issue

Map generation fails → Keep old maps, report error, suggest retry

User doesn't respond → Default to "skip" for temp files


Critical Notes

Memory Preservation

This skill is the ONLY way to:

  • Create historical session reports
  • Clean temporary files from workspace

Context Persistence

Without this skill running at session end:

  • Progress updates are lost
  • Navigation maps become stale
  • No historical record is created
  • Temp files accumulate

Never skip this skill - it's the foundation of context preservation!

Workflow Philosophy

This skill embodies the Nexus philosophy:

  • Memory preservation: Nothing is ever lost
  • Context awareness: Full system state captured
  • Progressive disclosure: Load what you need, when you need it
  • User collaboration: Interactive choices for important decisions

Resources

references/

  • workflow.md: Complete 9-step workflow (with TOC)
  • error-handling.md: All error scenarios and solutions

Integration with bulk-complete Skill

This skill uses the bulk-complete system skill for efficient task completion:

Step 2 & 2.5: Auto-runs bulk-complete when project work is done

python 00-system/skills/bulk-complete/scripts/bulk-complete.py --project [ID] --all --no-confirm

See bulk-complete/SKILL.md for standalone usage, all options, and test coverage details.


Remember: This is THE most important system skill. Every session MUST end with close-session to preserve context!

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

04

需要参考平台分布和安装热度时

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

补充不同宿主或平台的使用分布数据

能力 5

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

Claude Code

30.34%
按下载量换算26

Antigravity

20.09%
按下载量换算17

windsurf

15.51%
按下载量换算13

Codex

12.55%
按下载量换算11

OpenCode

7.31%
按下载量换算6

Gemini CLI

3.55%
按下载量换算3

安全审计

Gen Agent Trust Hub

未通过

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。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。

来源信息

继续浏览同类 Skills