apify-cli-guide

Static two-page guide, deployed as a Cloudflare Worker (Assets), served at https://guide.organizedai.vip/apify-cli and https://guide.organizedai.vip/apify-cli/workshop.

It also includes the Agent Session Miner Apify Actor, served at https://guide.organizedai.vip/apify-cli/workshop/agent-session-miner.

Deploy (copy-paste)

npm install -g wrangler   # skip if already installed
wrangler login             # skip if already logged in
wrangler deploy

That's it. Wrangler reads wrangler.toml, uploads whatever changed in public/, and prints the live URL.

Verify

curl -s -o /dev/null -w "root:     %{http_code}\n" https://apify-cli-guide.jordan-691.workers.dev
curl -s -o /dev/null -w "workshop: %{http_code}\n" https://apify-cli-guide.jordan-691.workers.dev/workshop
curl -s -o /dev/null -w "miner:    %{http_code}\n" https://apify-cli-guide.jordan-691.workers.dev/workshop/agent-session-miner
curl -s -o /dev/null -w "vanity:   %{http_code}\n" https://guide.organizedai.vip/apify-cli
curl -s -o /dev/null -w "vanity:   %{http_code}\n" https://guide.organizedai.vip/apify-cli/workshop
curl -s -o /dev/null -w "vanity:   %{http_code}\n" https://guide.organizedai.vip/apify-cli/workshop/agent-session-miner

All six should print 200.

Agent Session Miner

The Actor lives in claude-log-miner/. It scans Codex and Claude Code JSONL session logs, extracts session taxonomy records, ranks custom plugin opportunities from evidence in those logs, and generates an interactive HTML report with checkboxes for Codex and Claude Code import plans. The generated opportunities are not a fixed starter list; local runs derive names, descriptions, ratings, and evidence from the user's own session data.

Run the whole local-to-Worker flow:

cd claude-log-miner
npm install
npm run doctor
apify run --purge --input-file ./local-input.json
npm run export:worker-assets
cd ..
wrangler deploy

Use npm run doctor:strict before cloud operations when you want auth checks for both Apify and Cloudflare. It requires apify login and wrangler login.

For direct local imports:

cd claude-log-miner
npm run report

Open http://localhost:4177, select the plugin opportunities to keep, choose Codex, Claude Code, or both, then click Import Selected. On the deployed static Cloudflare page, browser filesystem restrictions mean the same button downloads an import plan and shows a command to apply it locally.

Add a new page

Drop a new .html file into public/, keep the same <style> block for a matching look, then run wrangler deploy again. No config changes needed — guide.organizedai.vip/apify-cli/<filename> picks it up automatically once the Worker is live (the vanity router forwards any sub-path straight through).

Structure

apify-cli-guide/
  claude-log-miner/  # Apify Actor: Agent Session Miner
  public/
    index.html                         # guide.organizedai.vip/apify-cli
    workshop.html                      # guide.organizedai.vip/apify-cli/workshop
    workshop/agent-session-miner.html  # guide.organizedai.vip/apify-cli/workshop/agent-session-miner
  wrangler.toml

---

For AI coding agents

If you were handed only this repo's URL, do this, in order, with no clarification needed:

1. Clone the repo and cd into it.
2. If the task is about the Agent Session Miner, cd claude-log-miner, run npm install, then run npm run doctor.
3. To regenerate the miner report, run apify run --purge --input-file ./local-input.json, then npm run export:worker-assets, then cd ...
4. Check wrangler --version. If missing: npm install -g wrangler.
5. Check wrangler whoami. If not logged in: wrangler login (this opens a browser — tell the user to complete it, then continue).
6. Run wrangler deploy from the repo root (no flags needed — everything is in wrangler.toml).
7. Run the six curl checks in Verify above. All must return 200.
8. Report the live URLs back to the user. Do not modify wrangler.toml (name, compatibility_date, [assets] block) — those are load-bearing for the existing Cloudflare Worker and vanity-domain routing.

If step 7 fails on the guide.organizedai.vip checks but the matching workers.dev checks pass: the Worker deployed fine, but the vanity subdomain routing is a separate, unrelated Cloudflare Worker (organizedai-vanity-router) that this repo does not own or deploy — report that distinction rather than trying to fix it here.