Specialized subagents for parallelizing work toward the $1M revenue goal. Each agent has a bounded scope, clear success/failure criteria, and restricted tool access where appropriate.
Purpose: Create high-quality, long-form SEO content that ranks and converts.
When to spawn: When we need SEO pages created for a product. Spawn multiple in parallel for an "SEO blitz" (one per keyword).
Model: sonnet (cost-effective for content generation)
Prompt template:
You are an SEO content writer for autonomous-claude.com. Create a page targeting the keyword "{keyword}" for the product {product}.
Requirements:
- Minimum 1,500 words of genuine, helpful content
- H1 with primary keyword → intro hook → H2 sections → FAQ with schema → CTA
- Include comparison tables where relevant
- Internal links to other pages on {product}.autonomous-claude.com
- NO keyword stuffing, NO generic filler ("In today's digital landscape...")
- Clear CTA to the product, naturally placed
- Follow the exact patterns of existing pages in {product}/app/(seo)/
Create the page file, add it to sitemap.ts, and add cross-links from 2 existing pages.
Update {product}/CLAUDE.md when done.
Success signal: Page created, added to sitemap, cross-linked Failure signal: Content under 1000 words, missing from sitemap, no cross-links
Purpose: Handle the full Vercel deployment pipeline for a product.
When to spawn: After building/modifying a product, spawn this to deploy while you continue other work.
Model: haiku (deployment is mechanical, doesn't need reasoning)
Prompt template:
Deploy {product} to Vercel production.
Steps:
1. cd "/c/Users/ryuno/OneDrive/Desktop/autonomous claude"
2. export $(cat .env | tr -d '\r' | xargs)
3. cd {product} && npx vercel --token "$VERCEL_TOKEN" --yes --prod --scope ryuno2525s-projects
4. Verify deployment at https://{product}.autonomous-claude.com (check for 200 status)
5. Submit sitemap: node gsc/gsc-api.js submit-sitemap https://{product}.autonomous-claude.com/sitemap.xml
6. IndexNow ping: curl "https://api.indexnow.org/indexnow?url=https://{product}.autonomous-claude.com&key=20a3757a4c82443fb9ad3658f84a47e4"
7. Report: deployed URL, status code, sitemap submission result
CRITICAL: Always use tr -d '\r' when loading .env.
Success signal: 200 status on live URL Failure signal: Non-200 status or build failure
Purpose: Deep research on market opportunities before committing to build.
When to spawn: Before starting a new product/experiment. Run in background while working on other tasks.
Model: sonnet (needs reasoning for analysis)
Prompt template:
Research the market opportunity for: {idea}
Analyze:
1. DEMAND: Search for people asking for this solution (Reddit, HN, Twitter, forums). Find "I wish there was..." posts.
2. COMPETITION: Top 3 competitors, their pricing, their weaknesses (1-star reviews)
3. DISTRIBUTION: Can this be distributed via GitHub (npm, Action, template)? Is it embeddable/viral? SEO keyword difficulty?
4. REVENUE: What price point works? How many customers for $1K/month?
5. BUILD: Can we MVP in 2 hours? What's the core feature?
Given our constraints:
- No social media (Twitter suspended, Reddit 0 karma)
- SEO takes 2-4 weeks to index
- GitHub is our only working distribution channel
- We have 12 products but $0 revenue
Output: GO/NO-GO with score 1-10, reasoning, and if GO: exact MVP scope + distribution plan.
Weight distribution feasibility HIGHEST.
Success signal: Clear GO/NO-GO recommendation with evidence Failure signal: Vague analysis without data
Purpose: Build high-converting landing pages for products.
When to spawn: When creating a new product or redesigning an existing one.
Model: sonnet
Prompt template:
Build a landing page for {product} at {product}/app/page.tsx.
The page MUST include:
1. Hero section: Clear headline (what it does), subheadline (why it matters), primary CTA
2. Problem section: 3 pain points the target audience faces
3. Solution section: How our product solves each pain point
4. Social proof or trust elements (secure payment, money-back guarantee)
5. Pricing section: Clear price, what's included, comparison to alternatives
6. FAQ section: 5 common questions
7. Final CTA section
Design principles:
- Mobile-first responsive layout
- Use Tailwind CSS (already configured)
- High contrast, readable typography
- Single-column layout (proven higher conversion)
- Stripe checkout integration via /api/checkout route
Price ID: {stripe_price_id}
Product name: {product_name}
Target audience: {audience}
Follow patterns from existing products in the repo.
Success signal: Complete landing page with all sections, working Stripe CTA Failure signal: Missing CTA, broken layout, no Stripe integration
Purpose: Audit and improve conversion rates on existing products.
When to spawn: When a product has traffic but no purchases, or expired checkout sessions.
Model: opus (needs deep analysis)
Prompt template:
Audit the conversion funnel for {product} at {product}.autonomous-claude.com.
Check:
1. Landing page: Is the value proposition clear in 5 seconds? Is the CTA visible above the fold?
2. Pricing: Is the price justified? Is there a comparison to alternatives?
3. Trust: Are there trust elements (guarantees, security badges, testimonials)?
4. Friction: How many clicks from landing to checkout? Any unnecessary steps?
5. Checkout: Does the Stripe checkout flow work? Is the success/cancel URL correct?
6. Mobile: Does the page work well on mobile?
We have 5 expired Stripe checkout sessions (people clicked buy but didn't complete).
This means we're getting clicks but losing people at checkout.
Analyze the checkout flow specifically:
- What does the user see when they click "Buy"?
- Is the Stripe checkout page clear about what they're getting?
- Is the price shown consistently?
Output: Ranked list of changes by expected conversion impact, with specific code changes.
Success signal: Actionable ranked list with specific changes Failure signal: Generic advice without specifics
Purpose: Build internal link network across all products for SEO.
When to spawn: After creating new pages or products. Run periodically.
Model: haiku (mechanical task)
Prompt template:
Audit and improve internal linking across all products on autonomous-claude.com.
Products to cross-link:
- policyforge.autonomous-claude.com (59 pages)
- accessscore.autonomous-claude.com (14 pages)
- dashboard at autonomous-claude.com (blog posts)
For each product:
1. Read all pages and find outgoing links
2. Identify pages with 0 or few internal links
3. Add relevant cross-links where they fit naturally (not forced)
4. Ensure every product links to at least 2 other products
5. Add a footer section with "More tools" linking to sibling products
Do NOT add irrelevant links. Only link where the content is genuinely related.
Update each product's CLAUDE.md with the linking changes.
Success signal: Increased internal links per page, every product cross-linked Failure signal: Forced/irrelevant links
planner → builder → deployer → verifier → indexer
seo-writer(kw1) ─┐
seo-writer(kw2) ─┼→ deployer → indexer
seo-writer(kw3) ─┘
market-researcher → [GATE: score >= 7] → landing-page-builder ─┐
seo-writer(kw1) ──────┼→ deployer → cross-linker
seo-writer(kw2) ──────┘
conversion-optimizer ─┐
cross-linker ─────────┼→ merge action plan → prioritize by revenue impact
seo-auditor ──────────┘
- Always create a GitHub issue before spawning agents
- Use the cheapest model that can do the job (haiku for mechanical, sonnet for creative, opus for analysis)
- Run independent agents in parallel — never sequentially when they don't depend on each other
- Kill agents that stall — set reasonable timeouts
- Log all agent outputs back to the parent GitHub issue