Tech Discussion Manager

技术讨论管理技能，自动创建结构化的讨论记录、代码输出目录，沉淀技术决策和创意，支持多智能体协同使用。

Security Scan

The skill mostly does what it says (create templates and directories) but it also modifies agent-wide rules (AGENTS.md) and references an undeclared workspace env var, which is more invasive than a simple template installer and deserves caution.

Purpose & Capability

Creating discussion templates, code-output directories, and index/archival workflows matches the skill description. However, the install step that appends mandatory behavior into AGENTS.md (forcing all technical discussions to use this skill and banning other recording locations) is stronger than a typical template installer — it's consistent with the stated goal of 'integrating with the memory system' but is heavy-handed.

Instruction Scope

Runtime instructions run three shell steps that create directories, copy templates, and append text into $OPENCLAW_WORKSPACE/AGENTS.md. The script references $OPENCLAW_WORKSPACE and writes to AGENTS.md and later describes writing to MEMORY.md and docs — the skill therefore instructs the agent to modify workspace files and agent rules. The instruction set also contains behavioral mandates (automatic triggering on keywords, enforced use) that go beyond pure templating and affect agent behavior broadly.

Install Mechanism

All install actions are local shell commands (mkdir, cp, grep/printf). No network downloads or external packages are pulled. This is low-risk from an install-origin perspective.

Credentials

The SKILL.md relies on the environment variable $OPENCLAW_WORKSPACE but the skill metadata declares no required env vars. That mismatch is a problem: the scripts will act on whatever path that env var points to (or fail if unset). The skill requests write access to workspace files (AGENTS.md, MEMORY.md) without declaring or explaining why these agent-level files should be modified permanently.

Persistence & Privilege

The install step appends persistent rules into AGENTS.md that change agent behavior (mandating use of this skill and disallowing other recording locations). Although always:false, this is effectively granting the skill persistent influence over agent policies and other agents' behavior, which is a privileged action and should require explicit admin consent.

What to consider before installing

This skill is not obviously malicious, but it performs permanent changes to your agent workspace and policy files. Before installing: 1) Back up $OPENCLAW_WORKSPACE/AGENTS.md and MEMORY.md. 2) Confirm what OPENCLAW_WORKSPACE is set to in your environment (the SKILL.md uses this env var but the skill metadata doesn't declare it). 3) Review the exact text the skill will append to AGENTS.md and decide whether you want a skill to enforce mandatory behavior across agents. 4) If you prefer more control, run the three shell steps manually (mkdir, cp, and append) after reviewing templates, instead of letting the skill modify AGENTS.md automatically. 5) If you install, monitor for unexpected changes to other agent configuration files. If you want, I can produce a safe, non-invasive installation script that only creates the directories and copies templates without altering AGENTS.md.

tech-discussion-manager 技能

功能概述

自动管理技术讨论和创意沉淀的标准化技能，帮助智能体自动创建规范的目录结构、讨论记录模板，确保技术知识完整留存。

核心功能

• 📝 标准化技术讨论记录模板
• 📂 自动创建关联代码输出目录
• 🔄 自动维护讨论索引
• 📦 讨论归档机制
• 🤝 多智能体协同支持

智能体使用规范

当用户发起技术讨论时，智能体必须按照以下流程执行：

1. 讨论启动流程

当用户提到以下关键词时，自动触发技能：

• "我们讨论一下..."
• "技术方案评审"
• "需求评审"
• "架构设计讨论"
• "创意 brainstorm"
• "记录一下这个讨论"

自动执行以下操作：

我已经帮你创建了技术讨论记录：
📝 讨论文件：`memory/tech-discussions/YYYY-MM-DD-主题.md`
📂 代码目录：`code-output/[主题]/`

我们可以开始讨论了，我会实时记录要点和决策。

2. 讨论记录格式

必须按照以下结构记录：

## 讨论背景
[自动填充讨论背景]

## 讨论要点
- [ ] 要点1：xxx
- [ ] 要点2：xxx

## 决策结论
- 结论1：xxx，原因：xxx
- 结论2：xxx，原因：xxx

## 待办事项
- [ ] 任务1：xxx，负责人：xxx，截止时间：xxx

3. 讨论结束流程

当用户表示讨论结束时，自动执行：

1. 整理讨论内容，提炼核心结论
2. 更新讨论状态为"已完成"
3. 将重要结论同步到 MEMORY.md
4. 如需归档，移动到 archive/ 目录
5. 更新 index.md 索引

4. 目录结构规范

$OPENCLAW_WORKSPACE/
├── memory/
│   └── tech-discussions/
│       ├── archive/                # 归档历史讨论
│       ├── TEMPLATE.md             # 讨论模板
│       └── index.md                # 讨论索引
├── code-output/                    # 代码输出目录
│   └── [讨论主题]/                  # 每个讨论对应独立代码目录
└── docs/
    ├── decisions/                  # 正式决策文档
    └── architecture/               # 架构设计文档

记忆系统集成

技能安装后会自动将规则写入 AGENTS.md，成为智能体的永久行为准则：

• 所有技术讨论必须使用本技能管理
• 禁止私自在其他位置记录技术讨论
• 所有讨论必须结构化留存，便于后续检索

ClawHub 发布说明

本技能符合ClawHub标准规范，可以直接发布到clawhub.com：

# 登录ClawHub
clawhub login

# 发布技能
clawhub publish tech-discussion-manager --version 1.0.0

用户使用示例

用户：我们讨论一下GEO工具的架构设计
智能体：✅ 已创建技术讨论记录：
📝 讨论文件：memory/tech-discussions/2026-03-29-geo工具架构设计.md
📂 代码目录：code-output/geo工具架构设计/
请开始讨论，我会实时记录要点。