Skip to content
Get packs

Archify Diagram Tuning

Runs experiments to measure and improve how well Archify generates accurate, readable diagrams

Source Download skill
Repository
tt-a1i/archify
Stars
92,587
Licence
Apache-2.0
Files
9 files

What it does

This skill is a maintainer workflow for Archify, a tool where an AI agent turns a described system into a diagram. Once installed, the agent follows a structured process: it sets up experiment rounds, generates diagrams from fixed inputs, reads agent traces, renders screenshots, compares before and after results, and records decisions in a journal kept outside the repository. It is meant to judge the whole diagram and the authoring experience, not just pass automated checks.

Great for

  • Maintainers of the Archify project evaluating changes to layout, routing or validation rules
  • Contributors running controlled comparisons of diagram generation quality and speed
  • Developers who want a repeatable replay of benchmark diagrams against two versions of Archify

Skip it if

  • General users who only want a diagram of their own system; this is an internal tuning process and needs the Archify repository and its tooling
  • Anyone wanting to share drafts publicly without care, because the tuning home holds content from source projects that may be private

Try asking

  • “Run a tuning round for the architecture diagram type on my project and record the first-draft results”
  • “Replay the benchmark against my candidate Archify build and report which drafts changed or flipped”
  • “Collect the Devin CLI session history for round architecture-2 and summarize failures and finalize runs”
  • “Render screenshots of the current candidate diagrams for visual comparison”
Step-by-step guide: how to use Archify Diagram TuningInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 9 original files, the Apache-2.0 licence and an attribution note.

    archify-tuning.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip archify-tuning.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip archify-tuning.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Set ARCHIFY_TUNING_HOME to a private directory outside the repository and never commit it, since drafts may contain private source content
  • Start with existing evidence and fixed JSON replays before running new generation experiments
  • Keep the first draft and every failure visible; do not overwrite benchmark entries with repaired drafts

What's inside

  • references/experiment-design.md
  • references/trace-export.md
  • references/trace-signals.md
  • scripts/collect.mjs
  • scripts/lib.mjs
  • scripts/replay.mjs
  • scripts/round.mjs
  • scripts/shots.mjs
  • SKILL.md

Redistributed unmodified from tt-a1i/archify under the Apache-2.0 licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.