AgentSkillsCN

idea-clawdbot

启动后台Clawdbot会话，探索和分析商业创意。说‘Idea: [描述]’来触发。‘idea’技能的分支重写为使用sessions_spawn，而非claude CLI + tmux + telegram CLI。结果发送到当前聊天，而非已保存消息。无外部依赖。

SKILL.md
--- frontmatter
name: idea-clawdbot
description: "Launch background Clawdbot sessions to explore and analyze business ideas. Say 'Idea: [description]' to trigger. Fork of 'idea' skill rewritten to use sessions_spawn instead of claude CLI + tmux + telegram CLI. Results sent to current chat, not Saved Messages. Zero external dependencies."
metadata: {"clawdbot":{"emoji":"💡"}}

Idea Exploration Skill (Clawdbot Native)

Launch autonomous background sessions to explore business ideas in depth. Get market research, technical analysis, GTM strategy, and actionable recommendations—all using built-in Clawdbot features.

Quick Start

Trigger phrase: Say Idea: [description] and the assistant will:

  1. •Spawn a background sub-agent session using sessions_spawn
  2. •Research and analyze the idea comprehensively
  3. •Save results to ~/clawd/ideas/<slug>/research.md
  4. •Send the file + summary back to this Telegram chat

How It Works

code
User: "Idea: AI calendar assistant"
       ↓
┌─────────────────────────────────┐
│  1. Detect "Idea:" trigger      │
│  2. sessions_spawn background   │
│  3. Sub-agent researches        │
│  4. Writes research.md          │
│  5. Returns to main chat        │
│  6. Sends file + summary        │
└─────────────────────────────────┘

Prerequisites

  • •Clawdbot with sessions_spawn enabled
  • •No external CLIs needed (fully native)

AGENTS.md Integration

Add this to your AGENTS.md:

markdown
## Idea Exploration

**When user says "Idea: [description]":**

1. Extract the idea description
2. Create a slug from the idea (lowercase, hyphens)
3. Use `sessions_spawn` to launch a background research session:
   - **task**: Use the template from `skills/idea-clawdbot/templates/idea-exploration-prompt.md`
   - **label**: `idea-research-<slug>`
   - **cleanup**: keep (so we can review the session later)
4. Confirm: "🔬 Research started for: [idea]. I'll ping you when done (usually 3-5 minutes)."
5. When the sub-agent completes, send the research file to the chat

**Result handling:**
- Research saved to: `~/clawd/ideas/<slug>/research.md`
- Send file as document via Telegram
- Include brief summary of verdict (🟢/🟡/🟠/🔴)

Analysis Framework

The exploration covers:

  1. •Core Concept Analysis - Problem, assumptions, uniqueness
  2. •Market Research - Users, TAM/SAM/SOM, competitors
  3. •Technical Implementation - Stack, MVP scope, challenges
  4. •Business Model - Revenue, pricing, unit economics
  5. •Go-to-Market Strategy - Launch, acquisition, partnerships
  6. •Risks & Challenges - Technical, competitive, regulatory
  7. •Verdict & Recommendations - Clear yes/no with action plan

Verdict Types

  • •🟢 STRONG YES - Clear opportunity, pursue aggressively
  • •🟡 CONDITIONAL YES - Promising but needs validation
  • •🟠 PIVOT RECOMMENDED - Core insight good, execution needs work
  • •🔴 PASS - Too many red flags

Example Output

code
~/clawd/ideas/ai-calendar-assistant/
├── metadata.txt
├── research.md    # 400-500 line comprehensive analysis

Tips

  • •Ideas typically take 3-5 minutes to analyze
  • •Check session progress: clawdbot sessions list --kinds spawn
  • •Monitor sub-agent: clawdbot sessions history <session-key>
  • •Results come back to the same chat automatically

Template Variables

When spawning the sub-agent, replace these in the prompt template:

  • •{IDEA_DESCRIPTION}: The actual idea text
  • •{IDEA_SLUG}: URL-friendly version (e.g., "ai-powered-calendar")