Blog
Measured numbers on tokens, screenshots, and what your coding agent can actually read.
Why your AI agent can't see what it built
An AI agent's world is text — it can't see what its code renders unless a tool hands it an image, and even then it needs to know what to look for.
Token EconomicsWhy your Claude Max plan runs out faster than you expect
Claude usage limits are token-based, not message-based — long visual sessions compound fast, and the fixes cost nothing.
AI Coding WorkflowMCP Explained (For People Who Just Want More)
MCP is Anthropic's open standard letting an AI agent call real tools and data instead of only generating text, one server per connected system.
Token EconomicsWhat Is a Context Window? (And Why Filling It Hurts)
A context window is the total text a model holds per turn; filling a 200,000-token window can cost 40% of it before a fix even starts.
Token EconomicsWhat does one Claude Code session actually cost?
A line-item breakdown of a two-hour Claude Code session shows screenshots and recordings, not code, driving the biggest single cost.
Token EconomicsVideo, Screenshots or Words: The Real Cost to an AI
A worked comparison of what a frame-dumped video, screenshots, and spoken words each cost an AI coding agent to read, using Anthropic's formula.
AI Coding WorkflowThe vibe coding workflow: a working setup for solo builders
The vibe coding workflow is describe, generate, look, react — the look-and-react half burns the most time and gets almost no dedicated tooling.
AI Coding WorkflowVibe Coding Tools in 2026: The Honest Stack
The 2026 vibe-coding stack has four jobs, agents, voice input, review, and testing, with prices given only where they're actually verified.
Reviewing AI CodeA 12-Point Checklist for Verifying AI-Built UI
Twelve checks for what AI coding agents most reliably miss, why they miss each one, and the fastest way to trigger it to verify the fix.
AI Coding WorkflowScreen Recording for Claude Code: What Works in 2026
Four ways to show Claude Code your screen, compared on token cost and what each one loses on the way to the agent.
Reviewing AI CodeHow to review AI-generated code you didn't write
The order that works: data shape, then boundaries, then logic. Top-to-bottom reading misses what AI-generated code gets wrong — and what to grep for.
Reviewing AI CodeHow to review what your AI coding agent built
How to review AI-generated code: what the agent already checked, what only you can catch, and a loop that closes the gap.
Reviewing AI CodeThe plan → execute → verify loop, explained
Plan and execute are easy to automate. Verification is where the quality is — and it's the one step an AI agent can't fully do by itself.
AI Coding WorkflowCan you feed a Loom video to an AI coding agent?
Loom's AI reads the transcript, not the pixels — so a Loom link never gets your coding agent the screen you pointed at.
Reviewing AI CodeHow to Give Feedback to Claude Code That Lands
Good feedback to Claude Code names the location, the expectation, and the gap between them — most reports name only the gap, and the fix misses.
Reviewing AI CodeHow to describe a visual bug to an AI coding agent
'The spacing looks off' gives an agent nothing to fix. Here's what to say instead, with a translation table of real vague-to-specific examples.
Token Economics11 Ways to Cut Claude Code Token Costs
Eleven ways to cut Claude Code token costs, ranked by tokens saved per minute of effort. Ten are free; one costs money.
Token EconomicsClaude Code token costs: where your context actually goes
A 1080p screenshot costs 2,691 tokens by Anthropic's own formula. Here's where the rest of a Claude Code session's tokens actually go.
AI Coding WorkflowBest Tools for Giving AI Coding Agents Feedback (2026)
A fair, priced comparison of Walkie, Clipy, Cobalt Capture, Vibeshots, Vibe Annotations, talkthrough-mcp, Jam.dev and plain screenshots.
Token EconomicsWhy Your Coding Agent Can't Watch a Screen Recording
Claude's API accepts images, not video. What actually happens when you frame-dump a screen recording for your agent — and the token cost.
ProductWhat Is a REVIEW.md Bundle? (Why Agents Read It, Not Video)
A plain look at Walkie's REVIEW.md bundle format — what's inside it, how big it is, and why your coding agent can read it but not your screen recording.
ComparisonsWalkie vs. Clipy vs. Loom: Which Tool Saves More Tokens?
A fact-checked comparison of Walkie, Clipy, Loom, and other screen-to-AI review tools on price, tokens, and features.
Token EconomicsHow Many Tokens Does a Screenshot Cost Your AI Coding Agent?
A measured, sourced breakdown of screenshot token cost for Claude and other coding agents, plus what to send instead.
AI Coding WorkflowGive Your AI Agent Visual Feedback Without Burning Tokens
The cheap way to show Claude Code or Cursor a UI bug: narrate and point instead of pasting screenshots. Here's the token math and the workflow.
Token EconomicsHow Claude Calculates Image Tokens (the Formula, Explained)
The exact formula Claude uses to price an image in tokens, worked through on a real 1920x1080 screen and a measured test screenshot.