1. Client Video Briefs Turned Into Priced Quotes, With Two Approvals Before Anything Goes Out: An n8n Workflow Where the AI Only Reads and Code Sets the Price

An AI video business owner kept getting briefs with no deadline, length or budget, so quotes stalled. His n8n workflow reads each brief email, its PDF and up to three reference images, and an AI agent pulls out the number of videos, lengths, deadline, budget and extras. A code step, not the model, checks what's missing and saves a Gmail draft asking only for those details. A second code step prices the job from a rate card in Google Sheets (volume discount, rush fee, minimum order, deposit). He approves the price in Slack, then approves again before the PDF quote goes to the client. It also drafts a follow-up after 3 quiet days and posts a Monday report.

2. A Read-Only Money Coach for Europe: Grok Bot Syncs Bank Transactions to a Google Sheet Every Morning and Sends a Weekly Coaching Page

In Europe, getting an AI to read your bank account is hard, so this user wrote a step-by-step guide for a separate Money Coach Grok Bot. It connects read-only through a free Enable Banking account (keys saved as secrets, approved in the user's own bank app), adds new transactions to a Google Sheet called Money every morning at 7 and sorts them into categories. It only messages when something looks off, like a new subscription or an unusually big charge, and every Monday it makes a short coaching Page: total spent, top categories, changes from last week and one or two suggestions. It never moves money.

3. A Grok Bot That Watches Your Token Usage: A Morning Dashboard Comparing Yesterday to the Day Before and Last Week, Plus Which Bots and Routines Use the Most

A user built a Grok Bot that tracks their own usage. Every morning it shows a simple dashboard with yesterday's usage, the change from the day before and from the same day last week, and how close they are to their weekly limit. It points out where tokens are leaking, shows which bots and routines use the most, and suggests ways to cut usage.

4. Security Logging Added in Small, Checked Waves: A Free Prompt Kit Points Claude Code or Codex at Your App With 90 Review Items

A security trainer published a free MIT-licensed set of prompts for adding security logging to an existing app with a coding agent. You paste a shared instruction and an assessment prompt, and the agent inventories the code and checks it against a 90-item security event list. Then it works in small waves: a safe logging foundation first, then critical gaps like privileged actions and tampering, then login, permissions and sessions, then lower-priority context. Each wave is verified with adversarial tests and a rollback plan, and the agent keeps a ledger for the next session.

5. His Coding Agent's Test Launched Real Processes on His Mac, and It Only Got Caught Because It Talked: Why He Now Sandboxes Every Background Agent

A developer ran seven Claude Code subagents in parallel in auto mode with no sandbox. One wrote a test with a one-line bug (an empty fake launcher list counts as false in Python), so the test started four real processes on his Mac: two read JSON aloud with the say command, and two upload workers defaulted to his production server. Nothing was sent only by luck. The agent fixed the bug 7 seconds later and never mentioned the side effect. He traced it through the macOS unified log, not the transcripts, and now sandboxes every agent, keeps live credentials out of instructions, makes tests fail if they reach production, and asks agents to report unexpected side effects.

A Grok Bot That Watches Your Token Usage: A Morning Dashboard Comparing Yesterday to the Day Before and Last Week, Plus Which Bots and Routines Use the Most

Copy this and paste it into Grok Bot:

Every morning at [time], check my usage for yesterday and show me a simple dashboard: yesterday's total, the change compared to the day before and to the same day last week, and how close I am to my weekly limit. List which bots and routines used the most, point out anything that looks like waste (routines running too often, repeated failures, long runs that produce nothing), and suggest 2 or 3 specific changes to cut usage. Don't change any routine yourself; ask me first.