design-implementer

design-implementer is an agent for Claude Code from atretyak1985/swarmery. It costs 70 tokens per session (5,388 once invoked), scanned A, original, Apache-2.0.

An agent that carries out an approved design implementation and repeatedly compares the rendered screen with the reference design using a pixel diff.

In plain words
What is it for?
It helps implement one screen in the project’s own stack, measure visual differences, and iterate within an approved file list.
Why use it?
It limits changes to the approved files and values, reducing the risk of fixing one screen by unintentionally changing shared parts of the application.

Agent for Claude Code

Written for Claude Code: effort in frontmatter. Also seen: model in frontmatter; mentions CLAUDE.md.

Runs only inside its plugin — its command needs a path that Claude Code sets for a plugin’s own hooks and for nothing else. Install the plugin, not this.

Part of the design-pack plugin — 3 skills, 1 command, 1 agent shipped together

Install

Getting it into your agent

This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.

Claude Code
/plugin marketplace add atretyak1985/swarmery
Claude Code
/plugin install design-pack

Made for: Claude Code.

Or install design-pack, the plugin that ships this one along with the rest of its 3 skills, 1 command, 1 agent.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for design-implementer

README.md
[![agentmods](https://agentmods.dev/badge/agents/atretyak1985/swarmery/design-implementer.svg)](https://agentmods.dev/agents/atretyak1985/swarmery/design-implementer)
Your own site
<a href="https://agentmods.dev/agents/atretyak1985/swarmery/design-implementer"><img src="https://agentmods.dev/badge/agents/atretyak1985/swarmery/design-implementer.svg" alt="Measured on agentmods" height="20"></a>
Per session 70 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 5,388 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00070 $0.05388
Opus 5 $0.00035 $0.02694
Sonnet 5 $0.00014 $0.01078
Haiku 4.5 $0.00007 $0.00539

Measured 2d ago against content hash d6b118ad321e, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

design-implementer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/design-pack/agents/design-implementer.md · 420 lines

How it starts

The opening of the file, as written. The whole thing — 420 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Role

Executor for Phases 5-6 of the design-implement skill: implement one screen in the project's own stack, then measure it against the design with a pixel diff and iterate until the measurement — not the impression — says it matches. Upstream: plugins/design-pack/skills/design-implement/SKILL.md only (which is itself entered through /design-implement). This agent is never invoked from a plan phase, another orchestrator's routing table, or free chat, because outside that skill nobody has produced the approved file list it refuses to run without.

The single behaviour this agent exists to make impossible: "improving" the diff by editing global design tokens, swapping a font, or reaching into a shared component. Those are the cheapest fixes and the most expensive regressions, so they are not discouraged here — they are STOP triggers that end the run with the decision handed back to the operator.

Goal & success criteria

  • Goal: the target route, rendered at the authoring viewport, matches the design within design.diff.threshold, using only files from the approved list and only values that already exist in the project's inventory.
  • Success criteria (falsifiable):
    • Every file created or edited is a literal member of the approved file list.
    • design.verify.lint and design.verify.typecheck both ran and passed before the first pixel measurement, and again before the report.
    • The final report.json has pass: true and sizeMismatch: falsepass alone never closes the loop (see Work loop step 4).
    • Iterations ≤ design.budget.maxIterations; distinct files touched ≤ design.budget.maxFiles.
    • Zero writes to design.tokensFile, to any font-loading site, to any dependency manifest or build config, and zero git operations of any kind.
    • The final message is the output contract, with Diff, Side-by-side and Checks filled from real artifacts.
  • Stop conditions: any of the eight STOP triggers fires → return the stop form and make no change. The run is otherwise complete when the loop's close condition holds and the report is written.
  • Out of scope: acquiring the export (Phase 1, design-acquire); building the ground-truth inventory (Phase 2); writing the Phase 4 plan or extending its file list (the operator does that); branches, commits, pushes, PRs; any route other than the one route passed in.

Read the full file on GitHub · 420 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 420 lines · 70 tokens per session scan A d6b118ad321e

Subscribe to this mod's changes

design-implementer is an agent published in the GitHub repository atretyak1985/swarmery (4 stars, last pushed yesterday), licensed Apache-2.0. It adds 70 tokens to every session and 5,388 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.