▍ howsmyaicoding

powered by Header · the optimization layer for coding agents

How's your
AI coding setup?

One skill grades your agent's setup and tells you how your AI coding is really going based on your sessions.

$ curl -fsSL https://howsmyaicoding.com/install | sh

✓ MIT · ✓ code stays local · ✓ no account

runs anywhere your agent has a shell: Claude Code · OpenAI Codex · Cursor · Aider · Goose · +50 more

01THE LOOP

LLM agents are non-deterministic. You optimize one the way you'd optimize any such system: measure, enforce, prove.

audit
Local, read-only. Grades your config, spend, and supply-chain risk — then mines your sessions for where the work breaks down.
guardrails
A CLAUDE.md rule can be forgotten; a pre-commit gate and a test ratchet can't. Header installs the ones you're missing, and /header wrapup writes each session's learnings into committed memory.
experiments
A variant costs tokens, not hours. Header A/Bs uncertain changes on your own tasks and merges only the statistically significant wins.

02HOW'S YOUR AI CODING, REALLY?

Static config can't tell you where the work actually breaks down. Header reads your real sessions and pinpoints it — every finding comes with the move that ends it.

sessions · last 14 days · 38 transcripts
  • ! 7× branch-juggling → git worktrees
  • ! 3 fixes, no test run → a pre-commit gate
  • ~ plan-mode on 12% of multi-file work → plan first
  • ! same gotcha hit 4× → capture to memory

03SHOULD YOU MOVE TO A NEWER MODEL?

Ask /header opus-4.8 or /header fable-5. Header grounds the call in your own engine and spend, names the trade, then hands off to an A/B.

/header opus-4.8
  • · on claude-opus-4-7 @xhigh
  • → Opus 4.8 same price · the in-tier upgrade
  • → Fable 5 capability jump · 2× per-token
  • › verdict earned by an A/B on your tasks

04ADVICE TUNED TO YOUR REPO

A custom topic builds Header's live briefing — cited sources, fresh each audit — around the languages, frameworks, and tools your repo actually uses. Commit it to .header/config and the team inherits it.

The audit and public briefings are free forever. A custom topic in Header comes with 3 free briefings, then $15 a month.

topic · default (public)
  • · agentic-coding generic feed
  • · self-improving-agent generic feed
topic · matched to your repo
  • ✓ next.js 15 · drizzle · bun
  • ✓ your MCP servers
  • ✓ the model you ship on

Header for teams · enterprise

One grade per repo. Now roll up all of them.

See which teams ship on unpinned models, F-grade harnesses, or no pre-commit gate — fleet grades, committed policy, and private briefings, in one view.

› contact us for a demo

One command. Be honest about your setup.

$ curl -fsSL https://howsmyaicoding.com/install | sh