Learning path map
Detailed content
🔍 n8n Workflow Reviewer
A skill that has Claude review your n8n automations like a senior architect: five categories, findings named node by node, and a prioritized list of fixes.
Treat an n8n workflow as code: it has bugs, technical debt, and failure points. The reviewer performs the audit you would do in a pull request.
Automations break in production precisely where no one looked. A structured review catches that before the customer does.
Structured audit · actionable finding · senior engineer persona · "no fluff".
The skill accepts complete JSON, partial JSON, a text description, an error message, or a canvas screenshot—and adapts what it can review for each format.
Accepting imperfect input is what makes the skill truly usable—nobody always has clean JSON on hand.
Flexible input · graceful degradation · "say what can't be evaluated".
The fixed framework: 🔴 breaking errors, 🟡 silent errors, 🔵 performance, 🟢 maintainability, and ✅ priority list.
Running the same five categories every time removes reviewer bias and ensures complete coverage.
Fixed checklist · color-coded severity · "don't skip any category".
No Error Trigger, no retry, no timeout, no failure alert — the workflow “runs” but loses data without anyone noticing.
It's the most valuable category: the error that appears is easy to spot; what's missing is what destroys trust in the system.
Error Trigger · retry/timeout · alert on failure · "what data will disappear".
Unnecessary API calls, missing pagination, loops where batch processing would work, nodes with default names, and logic buried in expressions.
Hidden costs eat into the margin; bad names make the "me six months from now" lose hours. Both have concrete fixes.
Batch > loop · pagination · descriptive names · sticky notes.
The tone rules (be direct, name the node, give the exact fix), the output format for each category, and the final verdict with a score from 0 to 10.
It’s the template you adapt to review any technical artifact—not just n8n.
Direct tone · findings format · honest score · original SKILL.md.
🔥 Local Leads Abundance System
A skill pipeline that finds local businesses in directories, enriches contacts, creates a website as a gift, and starts a conversation through the contact form—all chained together.
The overview: each step is a skill, and the output of one is the input for the next—a pipeline, not one giant skill.
It's the canonical example of combining several small skills into a system that delivers end-to-end results.
Pipeline · contract between stages · one skill per stage.
Use browser automation to open each card in a directory, read the profile, and structure the data in a spreadsheet.
The lead source determines the quality of the entire funnel — directories pre-qualify established businesses.
Browser automation · prequalified source · structured data.
Try one provider; if it doesn't find anything, move on to the next. Chaining multiple sources finds websites and LinkedIn profiles with high accuracy.
No single source covers everything; the cascade maximizes the success rate without paying for all of them at once.
Waterfall · ordered fallback · increasing cost · accuracy.
For each lead, create an improved version of their site (clone or redesign) and send it as a gift to start a conversation.
It's the difference between generic cold outreach and an approach where you've already delivered value before asking for anything.
Lead magnet · upfront value · personalization at scale.
The final step automatically fills out and submits the business’s own website contact form, with the gift attached.
Leaving the saturated channel (cold email) changes the response rate—and the sending screenshot becomes proof of delivery.
Alternative channel · proof of delivery · ethical automation.
How an orchestration skill calls each stage, passes the CSV along, and saves progress so you can pick up where you left off.
A long pipeline needs checkpoints and a data contract, or an error in step 4 makes you repeat steps 1 to 3.
Orchestration · checkpoint · CSV contract · idempotency.
📊 Lead Scoring with Apify
A qualification skill that classifies each deal, scores brands across three dimensions using Google Search via Apify, and ranks them by outreach priority.
Assign a score to each lead to order the outreach queue by likelihood to convert, instead of contacting them at random.
The bottleneck in outreach is your time. Prioritizing leads is what multiplies the return on the same effort.
Prioritization · ICP · repeatable, unbiased scoring.
Before scoring, the skill reads the site and classifies the business—only businesses with their own product (brand) enters scoring.
Scoring someone who isn't the target is wasteful. The classification filter cuts the noise before using API calls.
Decision tree · signals in the HTML · filter before spending.
Each brand gets 0–50 points for product development capability, 0–30 for data maturity, and 0–20 for brand maturity.
Dimensions with explicit weights make the score easy to understand: you know why a lead scored highly, not just that it did.
Three dimensions · weights 50/30/20 · explainable score.
Pass 1 reads only the site (free); pass 2 runs 5 Google searches via Apify per brand to find LinkedIn job titles, openings, and awards.
Collecting the low-cost signal first and only then paying for the search is the pattern that keeps cost per lead low.
Two-pass · free vs. paid signal · ~$0,045 per brand.
The score becomes a tier (≥70 Tier 1, 50–69 Tier 2, 30–49 Tier 3, <30 Tier 4), and the CSV includes subscores and the evidence for each one.
A clear output contract lets the next skill (outreach) consume the result without guesswork.
Tiers · evidence by dimension · contract for downstream use.
Two modes (full run and --phase2-only) and a checkpoint file that saves after each milestone so you can safely resume.
Scoring 78 brands takes hours; without a checkpoint, an interruption costs repeated time and API money.
Checkpoint · idempotency · operating modes · author-created SKILL.md.