LeoYeAI/agent-master-skills / skills/acquire-first-1000-users-on-reddit
first-1000-users
VerifiedAI-powered Reddit seeding agent for founders. Analyzes a product spec, maps relevant subreddits, finds real threads where target users need help, drafts personalized replies and DMs, and posts approved outreach via Reddit API. Use when someone wants to find and engage their first users on Reddit, seed a product launch, or do community-led growth without a budget.
| File | Kind | Size |
|---|---|---|
| README.md | markdown | 255 B |
| SKILL.md | skill md | 16.5 KB |
| _meta.json | config | 310 B |
| first-1000-users-spec.md | markdown | 14.8 KB |
| first-1000-users-spec.v2.md | markdown | 22.2 KB |
| first-1000-users-spec_v3.1.md | markdown | 9.1 KB |
| processing-logic-v3.1.md | markdown | 7.0 KB |
first-1000-users-spec_v3.1.md
View on GitHub# SPEC DOCUMENT — `first-1000-users`
**An OpenClaw Skill for AI-Powered Reddit Seeding**
Platform: OpenClaw (ClawBot)
Version 3.1 | February 2026
Status: Draft — Pending Supervisor Approval
---
## 1. Overview
`first-1000-users` is an OpenClaw skill that helps founders seed their product into real Reddit conversations. The AI agent analyzes a product spec, maps relevant subreddits, finds real threads, drafts personalized replies and DMs, and executes approved outreach via Reddit API.
### 1.1 Core Idea
Find people who are **already looking** for what you built on Reddit. Show up with a helpful reply in the thread and a thoughtful DM to the person who raised the problem. AI handles research, discovery, and drafting. Human approves every message before it goes out.
### 1.2 Two Engagement Channels
| Channel | When to Use | Conversion |
|---------|------------|------------|
| **Public Reply** | Thread is active, multiple people have the problem | Lower per-reply, visible to many |
| **Direct Message** | One person clearly has the pain point | Higher per-message, 1-on-1 |
### 1.3 What This Skill Is
- ✅ AI agent: researches subreddits, finds real threads, drafts personalized outreach
- ✅ Executes approved outreach on Reddit via API / browser automation
- ✅ Human-in-the-loop: every message requires explicit approval
### 1.4 What This Skill Is NOT
- ❌ Not fully autonomous — every action requires human approval
- ❌ Not a spam tool — hard-coded rate limits and anti-abuse protections
- ❌ Not a fake-account system — user's real Reddit account only
### 1.5 Execution Model
```
AI does: Research, Search, Draft, Post (after approval), Monitor
Human does: Approve, Edit, Reject, Handle follow-up conversations
Rule: AI NEVER sends anything without explicit approval
```
---
## 2. How It Works — 6-Phase Pipeline
```
Phase 1: RESEARCH → Analyze spec, map subreddits, generate signals
Phase 2: DISCOVERY → Search Reddit for real matching threads
Phase 3: DRAFT → Personalized reply/DM per thread
Phase 4: APPROVE → Human reviews each draft [GATE]
Phase 5: EXECUTE → Post approved messages
Phase 6: MONITOR → Track engagement, alert on responses
```
### 2.1 Input
A `.md` file describing the product:
| Section | What to Include |
|---------|----------------|
| Product name & one-liner | What it is in one sentence |
| Problem it solves | Pain point in user's language |
| Who it's for | Target audience (specific) |
| Key features | Top 3–5 differentiators |
| Pricing | Free / freemium / paid |
| Current stage | Pre-launch / beta / live |
| Product URL | Link or "not yet" |
| Competitors | What people use today instead |
### 2.2 Outputs per Phase
| Phase | Output |
|-------|--------|
| 1. Research | Subreddit Map, Buying Signals, Style Guide |
| 2. Discovery | Ranked Thread Queue (real threads) |
| 3. Draft | Personalized messages per thread |
| 4. Approve | Human-vetted message queue |
| 5. Execute | Action log with statuses |
| 6. Monitor | Engagement report |
---
## 3. Phase 1: Research
### 3.1 Subreddit Map
Ranked list of subreddits, verified via browser/API:
- Verified size, relevance score, self-promo rules, DM culture, entry strategy
- Scored on 5 axes: problem discussed, audience present, activity, tool-friendly, DM-receptive
- Only include 3+/5 scores
- Target: **5–8 subreddits**
### 3.2 Buying Signal Library
Categorized search phrases, specific to the product:
- 5 categories: Direct Request → Comparison → Pain Point → Workflow Question → Discussion
- Each signal includes: pattern, Reddit search query, real example, engagement channel, recency filter
- At least 4 signals per category
### 3.3 Style Guide
Derived variables: OFFER_TYPE, MAKER_FRAMING, SWITCHING_COST, tone notes.
---
## 4. Phase 2: Discovery
Search Reddit for real threads matching signals. Filter by recency, engagement, competition. Score 0-10. Present ranked Thread Queue to user.
Max 50 threads per session. Refresh daily.
---
## 5. Phase 3: Draft
Read full thread → draft personalized message (not template fill-in).
**Reply structure:** Acknowledge → Help → Bridge → Soft close
**DM structure:** Reference → Empathize → Value → Introduce → Low-pressure close
Automated quality gates run before presenting to user.
**Tone:** Lowercase "i", no em dashes, human filler, messy numbers, product mention at end. 3-6 sentences for replies, 3-4 max for DMs.
---
## 6. Phase 4: Approve
Human reviews each draft: Approve / Edit / Reject / Skip. Final confirmation before execution.
---
## 7. Phase 5: Execute
Post via Reddit API or browser. Rate limits enforced:
| Action | Limit |
|--------|-------|
| Replies | 5/hour |
| Same subreddit | 2/day, 2 min cooldown |
| DMs | 10/day, 5 min between |
| Per session | 20 max |
| Per day | 30 max |
**Safety triggers:** removal → pause 48h, warning → pause all 24h, ban → full stop, removal rate > 10% → full stop.
---
## 8. Phase 6: Monitor
Track responses every 30 min (first 24h), then daily. Alert on replies. Draft follow-ups (still require approval). Never auto-reply. Never follow up on unanswered DMs.
---
## 9. Ethical Guidelines
### 9.1 Hard-Coded Rules
- Human approval required for every message
- One DM per person (contacted_users database)
- Rate limits cannot be overridden
- No fake accounts
- Every message personalized per thread
- Bans permanently block community
- No follow-up DMs on no-reply
- Auto-pause on removals/warnings
### 9.2 Disclosure
User is the sender. AI is a drafting tool. User approves every message. Analogous to having an assistant draft messages. If Reddit rules change to require AI disclosure, skill should be updated.
---
## 10. Technical Requirements
### 10.1 Skill Structure
```
~/.openclaw/skills/first-1000-users/
├── SKILL.md # Main instructions (OpenClaw format)
├── scripts/
│ ├── reddit_search.py # Search for matching threads
│ ├── reddit_post.py # Post replies, send DMs
│ ├── reddit_verify.py # Verify subreddit status/rules
│ ├── engagement_tracker.py # Track responses
│ └── rate_limiter.py # Enforce rate limits
├── config/
│ ├── rate_limits.json
│ ├── approval_level.json
│ └── banned_list.json
├── data/
│ ├── product_spec.md
│ ├── subreddit_map.json
│ ├── signals.json
│ ├── style_guide.json
│ ├── thread_queue.json
│ ├── drafts/
│ ├── approved/
│ ├── sent/
│ └── contacted_users.json
└── logs/
├── actions.log
├── errors.log
└── metrics.log
```
### 10.2 Dependencies
```yaml
requires:
env: [REDDIT_CLIENT_ID, REDDIT_CLIENT_SECRET, REDDIT_USERNAME, REDDIT_PASSWORD]
bins: [python3, playwright-cli]
python: [praw, playwright]
```
---
## 11. Build Plan — 2-Day Sprint
### Day 1: Research + Discovery + Draft (Phase 1-3)
**Morning:**
- Set up OpenClaw skill directory and SKILL.md with YAML frontmatter
- Implement Step 0: spec parsing, variable extraction, validation
- Implement Phase 1: subreddit mapping with live verification, signal generation, style guide
- Write `reddit_verify.py` for subreddit rule/status checking
- Test: feed product spec → verify research output quality
**Afternoon:**
- Write `reddit_search.py` for thread discovery using signal phrases
- Implement thread scoring (signal_match + freshness + engagement + competition)
- Implement Phase 3: per-thread drafting with quality gates
- Implement approval presentation format
- Test: signal → real threads → personalized drafts → approval flow
**Day 1 Deliverable:** Working pipeline from spec input through draft approval.
### Day 2: Execute + Monitor + Safety (Phase 4-6)
**Morning:**
- Write `reddit_post.py` for posting replies and sending DMs
- Implement `rate_limiter.py` with hard limits
- Build contacted_users database for duplicate prevention
- Implement safety triggers (removal/ban detection, auto-pause)
- Implement error handling (429, 403, 404, network failures)
- Test: approved draft → execute → verify posted → log
**Afternoon:**
- Write `engagement_tracker.py` for response monitoring
- Implement follow-up draft suggestions (with approval gate)
- Full pipeline test: spec → research → discover → draft → approve → execute → monitor
- Test edge cases: thread locked, user deleted, rate limit hit, post removed
- Write documentation and example output
- Package and submit
**Day 2 Deliverable:** Complete skill with execution, monitoring, and safety.
---
## 12. Success Criteria
| Criteria | Target |
|----------|--------|
| Subreddit relevance | 4/5 top suggestions are active + on-topic |
| Signal accuracy | 5+ real threads per top 3 signals |
| Thread discovery | 10+ threads per session |
| Draft quality | 90%+ pass automated gates |
| Human approval rate | 70%+ approved without major edits |
| Execution success | 95%+ messages post successfully |
| Engagement | 30%+ reply response rate |
| Safety | 0 bans, < 5% removal rate in first 50 actions |
---
*End of Specification — first-1000-users v3.1 (Reddit-only)*
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |