Guides

A guide to installing Invoice Pack, connecting Gmail, and running it locally to separate matches, documents to review, and missing invoices.

The skills and prompts I use to plan before writing code, execute without losing direction, and then map and check what actually got built.

ManyChat charges per contact, so the bill grows with your audience. The same comment-to-DM automation, built from one signup and five Claude Code prompts.

Seven reference sources for flows and product screens, plus a prompt that turns inspiration into design instructions you can actually check.

One command shows what actually takes up the context window: conversation, memory, skills and MCP tools. Here is how to read the map and choose what to shorten, clean up or move to a separate agent.

Continue a local Claude Code session from your phone: setup, connection, safe usage, and troubleshooting, step by step with screenshots.

Out of space on your computer and deleting files does not help? A ready prompt for Claude Code that scans your machine, finds a runaway git repo eating your disk, and fixes it only with your approval.

Anthropic deleted more than 80% of Claude Code's system prompt for the new models, with no drop in code test results. Nine checks that go through your CLAUDE.md and tell you what to delete, what to keep, and what to move into a skill.

What your agent is allowed to see, what it should never see, and how to set it up right: the 1Password for Claude pattern, rules for working agents, and a compact checklist to copy.

Cowork looks like a chat, but it's built like an agent platform: an isolated Linux machine, a planner that breaks down tasks, and cloud sessions that keep running while your laptop is closed. The full map, plus three workflows that make use of it.

The ladder from Boris Cherny, creator of Claude Code: 5 levels of agent adoption, a 60-second self-check to find your level, and the one move that takes you up a level.

Mac version 1.24012.1 added Hebrew and Arabic files for the first time. I checked it on my Mac: Hebrew now displays correctly, right to left. Here's how to update, how to verify it, and what to do separately in VS Code.

How to check where a comment-to-guide funnel actually loses people. Perfect entry does not mean delivery. Four leak points and how to close each one.

The map of a content pipeline with 11 agents and two control gates: seven creators, two gates, a manager and a publisher, and a human eye at the end. How the work hands off between agents and where it stops.

A review agent that reads your diff, flags security weaknesses and stops the merge automatically on a serious finding. One skill, one gate on your code.

5 checks that separate data from instructions inside a tool-connected agent, with the two most important attacks CrowdStrike published and concrete mitigation.

What Claude Cowork is, how an agent that runs in the background and carries over between your computer and phone changes the shape of work, when to use it, and what matters about the beta.

A short checklist to run after every deploy: open the production URL itself and go route by route, because a green build does not prove production is alive.

Most people use Claude Code and Codex like a chatbot and miss the power. The move from a recommendation system to a worker system in 5 steps: install, connect tools, set limits, whole tasks, and daily improvement.

Let an agent gather scattered customer feedback, cluster it into themes, prioritize by pain, and propose concrete fixes. Includes a ready prompt and a weekly loop.

Install and use /watch for frames, captions, timestamped transcripts, and Whisper fallback when captions are unavailable.

A starter set for an agent that doesn't improvise: Superpowers, Caveman, Ponytail, Frontend Design, Design Review and Fullstack Eval.

A set of prompts that split work between Fable 5, a reading agent, a writing agent, a review agent and a progress file.

A short checklist that defines completion evidence before you let an agent work alone: tests, build, screenshot, and a review agent.

Splitting the work between the strong model and the fast model: planning, execution, review and fixing, with defined subagents and ready prompts for every stage.

When to start with Sonnet and when to open Opus in Claude Code, without burning a strong model on grunt work and without letting a fast model decide alone.

A short document that defines role, boundaries, permissions, evidence and a stop line before you let an agent work.

A short structure that turns an action that repeats twice into a workflow an agent can run again.

When to open a subagent, what to give it, and what to return to the main conversation without polluting the context.

A short checklist that separates read, propose, write and outbound action before an agent touches a real system.

How to pick a recurring process, clean it up, add variable inputs, and turn it into a procedure you can repeat.

A structure that separates stable memory from memory that goes stale, with source, date, scope and a refresh check.

A short structure that gives the agent boundaries, checks, stop conditions and a finish line before it touches the project.

How to split work between Cursor, Claude Code, Codex, subagents, and the human who owns product decisions.

Installation and usage guide for MemPalace, an open-source AI memory system that organizes memories like rooms in a palace. MCP setup, conversation mining, and tips.

Step-by-step guide to connect your Meta Ads account to Claude Code via MCP Server. Pull campaign data, analyze performance, and get recommendations.

How to manage CLAUDE.md, MEMORY.md, rules, skills and MCPs in Claude Code. A layered guide with an automated audit prompt for your setup.

Complete tutorial for building an automated knowledge map on any topic using parallel AI agents and D3 visualization.

Branch protection, required review and CI checks: the gate you need before letting background agents open a PR on their own.

Default to a cheap workhorse model, move up to an expensive one only for architecture, cross-system judgment, deep debugging and final review.

How you build authority for an agent one rung at a time: a small task, a limited action, repetition, a skill, and only then a scheduled agent.

Breaking down the /usage command by skill, subagent, plugin and MCP server, so you know what's burning through your quota before it runs out.

Five steps to roll out a code agent to the whole team, without burning the trust of developers who worry about being replaced or about broken code.

A short template for measuring time, failure rate and human hours before you turn an autonomous agent loose on an existing process.

A weekly radar that filters out the noise: Claude in Chrome, an automatic MCP connection fix, and a new dataviz skill.

Seven work templates that stop a loop from running forever, dying with the laptop, picking the wrong model, touching production, or declaring done too early.