| name | tool-directory-manager |
| description | Automates the lifecycle of the AI tools directory. It fetches new submissions from Google Sheets, summarizes them for approval, and triggers the screenshot engine for approved tools. |
Tool Directory Manager
This skill manages the intake and visual processing of new AI tools submitted to the directory.
Workflow
1. Fetch & Summarize Submissions
Run this to see what's new. It fetches the live CSV directly from Google Sheets.
curl -L "https://docs.google.com/spreadsheets/d/1VL4dAgyQK8EZLo6WqZeQ0zOmsCQcxQLhixmFtsJWC3A/export?format=csv" > latest_submissions.csv && python3 scripts/check_new_submissions.py
- Policy: You MUST present the summarized list of new tools to the user and ask for approval (or which ones to include) before proceeding to the next step.
2. Enrichment & Capture
For each approved tool, do not rely solely on the submission. You MUST visit the live site to extract high-quality data.
node scripts/enrich-batch.js
- Scraping Policy: Extract the meta description, headers, and body text to understand the real value proposition.
- Enrichment: Based on the scrape, generate:
- 4 Key Features: Actionable, benefit-driven bullet points.
- Pricing Details: Identify if it's Free, Freemium, or Paid with specific tiers if visible.
- Integrations: List specific platforms/APIs the tool works with.
- Visuals: Capture a high-res WebP screenshot locally to
public/screenshots/. NO Microlink required.
3. Merge & Update
Generate the TypeScript object for lib/ai-tools-data.ts. Ensure the slugify logic matches (dots to dashes).
4. Notify Submitters
After merging, notify the submitters that their tool is live.
RESEND_API_KEY=$(grep RESEND_API_KEY .env.local | cut -d'=' -f2 | tr -d '"') python3 docs/tool-directory-ops/send_emails.py
-
Email Copy Policy: Use a direct, conversational, and "human" tone. Acknowledge that the tool was submitted and that it has been handpicked for the directory section of Real AI Examples.
-
Approval: You MUST show the exact email copy to the user for approval before running the script.
-
Verification: Always confirm exactly which emails the script will send to before execution.
-
Cleanup: After a successful run, update the ALREADY_SENT list in docs/tool-directory-ops/send_emails.py with the new emails and push to main.
Tips
-
Always check the latest_submissions.csv timestamp to find tools submitted after the last update.
-
The site is primarily a library of practical AI blueprints; the directory is a section to feature handpicked tools that help people get practical work done.
-
If a tool name contains titles like "Dr" or "Coach," or if it's a University/Mega-Corp, flag it for the user as per the project's QC rules.
-
Never use "AI slop" or overhyped marketing language in communications.
