gtm-engine
npx machina-cli add skill PHY041/claude-agent-skills/gtm-engine --openclawGTM Engine — Composite Skill
Combines brand monitoring, lead generation, and outreach preparation into one automated GTM loop.
Architecture
brand-monitor → Tracks competitor mentions + buyer signals on Reddit
↓ (parallel)
lead-generation → Finds high-intent buyers across Twitter/Reddit/Instagram
↓
[merge + deduplicate signals]
↓
[score all leads 1-10]
↓
[prepare outreach drafts for warm leads ≥6]
↓
[send for human approval — NEVER auto-send]
Feedback Loop
Competitor Discovery: brand-monitor results feed back to lead-generation. If brand-monitor finds a new competitor mentioned (one not in the original config), it's automatically added to lead-generation's query set.
Step-by-Step
Phase 1: Competitive Intel (brand-monitor)
Call brand-monitor for all configured competitors.
Input → brand names from config
Output → alerts {subreddit, post_url, sentiment, intent, urgency}
Filter for buyer signals (intent = "buyer_signal" or competitor_comparison with negative sentiment toward competitor).
Phase 2: Lead Discovery (lead-generation) [PARALLEL with Phase 1]
Call lead-generation with product profile.
Input → product_url (auto-profile) + competitor names from config
Output → raw leads list {platform, username, post_text, url, posted_at}
Phase 3: Merge + Score
Combine Phase 1 buyer signals + Phase 2 raw leads.
Deduplicate by {platform}:{username}:{post_id} against data/lead-generation/sent-leads.json.
Score each using rubric (see lead-generation skill). Filter: only keep score ≥ 6.
Phase 4: Prepare Outreach
For each warm lead (score 6-7) and hot lead (score 8-10), draft a personalized outreach message:
- Warm: engage with their content first (like/reply)
- Hot: direct DM draft
NEVER send without human approval.
Phase 5: Report for Approval
🎯 GTM Engine — [date]
Competitive Intel:
- [N] buyer signals on Reddit
- Top: [subreddit] "[post title]" (score X)
Leads Found:
- 🔴 [N] Hot leads (score 8-10)
- 🟠 [N] Warm leads (score 6-7)
Top 3 Leads:
1. @username | [platform] | Score: [X]/10
"[post excerpt]"
Outreach: "[draft]"
Reply "approve [1,2,3]" to queue these for sending, or "skip" to discard.
I/O Contract Summary
| Phase | Skill Called | Key Input | Key Output |
|---|---|---|---|
| 1 | brand-monitor | competitor names | buyer_signals list |
| 2 | lead-generation | product_url | raw_leads list |
| 3 | (internal) | merged signals | scored_leads list |
| 4 | (internal) | scored_leads | outreach_drafts |
Source
git clone https://github.com/PHY041/claude-agent-skills/blob/main/composite/gtm-engine/SKILL.mdView on GitHub Overview
GTM Engine combines brand monitoring, lead generation, and outreach drafting into a single automated GTM loop for founders. It monitors competitors on Reddit, finds high-intent buyers across social platforms, and prepares warm outreach sequences that require human approval before sending.
How This Skill Works
It runs in five phases: Phase 1 brand-monitor gathers competitor signals from Reddit, Phase 2 lead-generation scouts high-intent buyers across Twitter, Reddit, and Instagram, then Phase 3 merges and scores the signals (1-10). Phase 4 prepares outreach drafts for leads scoring 6-10, with warm and hot variants, and Phase 5 reports for human approval before sending.
When to Use It
- You want competitive intel from Reddit to inform GTM positioning and messaging.
- You need a prioritized list of warm/hot leads for outreach across social platforms.
- You require a structured outreach pipeline that never auto-sends without approval.
- You want real-time signals on buyer intent and competitor mentions to guide GTM bets.
- You need an auditable flow that combines competitor intel, lead discovery, and outreach drafts with approvals.
Quick Start
- Step 1: Trigger the workflow with one of the commands: run gtm engine, find leads, competitive intel, or outreach pipeline.
- Step 2: Provide inputs (product_url) and optional competitors; let brand-monitor and lead-generation run in parallel.
- Step 3: Review the merged results, approve outreach drafts, and queue messages for sending.
Best Practices
- Configure the competitors list precisely; rely on auto-detection if omitted but review results.
- Use product_url to auto-profile for improved scoring and more relevant leads.
- Deduplicate signals by platform, username, and post to avoid duplicate outreach.
- Review outreach drafts in the approval stage and tailor messages before sending.
- Regularly refine scoring rubric and thresholds to reflect changing buyer signals.
Example Use Cases
- Competitor mentions surge on Reddit with buyer signals; GTM Engine surfaces 2 hot leads and 3 warm leads for review.
- Lead-generation returns multiple raw leads; scoring elevates 4 to 6+ and drafts are prepared for outreach.
- Outreach drafts are created but are not sent until a human approves them.
- Top lead includes a high-score user on a target platform with a personalized outreach draft.
- Weekly GTM digest summarizes competitive intel and the current outreach queue for leadership review.
