Memi

Design CI for coding agents.

Memi is the read-only design engineering audit and skill layer for coding agents. It gives Codex, Claude Code, Cursor, Grok Build, and MCP clients file-anchored UI evidence before merge.

The first pass reads the product you already have, identifies accessibility, hierarchy, state, responsive, motion, and token risks, then reruns the same deterministic check after a scoped fix. Your code remains the source of truth.

Supported today: Node 20, 22, and 24 on macOS, Linux, and Windows. Figma and Studio are optional companions.

npm · memoire.cv · current versions · MCP Registry · Agent Skills

Quickstart

Run one audit in any frontend repository:

npx -y @memi-design/cli@latest diagnose . --json --no-write --fail-on none

You get normalized finding IDs, confidence, provenance, and file:line evidence. No account, API key, Figma file, global install, or daemon is required.

Keep the workflow available to your coding agent:

npx skills add memi-design/memi --skill audit-frontend-design

Then ask: “Audit this frontend before editing it. Prioritize the five fixes that will matter most to users.”

Compatible with the shadcn registry and v0 design systems.

If Memi catches a real interface issue in your project, star the repository and share the finding. That is the most useful signal for deciding what to improve next.

Related MCP server: SIN-Code-Frontend-Design-Skill

Evidence at a glance

These are the measured results currently available in the V15 confirmatory audit. They describe that study; they are not estimates for your repository.

The full study also reports exclusions, failures, and limits. No superiority, speed, or dollar-savings claim is made.

Separate historical release record: the 2.7 candidate record reported 2,187 / 2,187 tests passed. It is release evidence, not part of V15 and not proof that every project benefits.

Benchmarks and paper

The graphic is a compact reading guide to the public technical paper, not a leaderboard. The primary measure was a blinded, model-graded 100-point design-quality rubric. The preregistered question was narrow: could Memi stay within five points of its paired baseline on each renderable task?

Benchmark contracts are separate from study results. InterfaceBench v1 specifies 100 target tasks with 5 pinned seed tasks; it is not an aggregate performance score. DesignWorkBench v2 holds 300 task contracts and still requires practitioner calibration before any certification claim.

What you get

The default workflow is deliberately read-only. Write-capable scaffolds and Figma operations are separate, explicit workflows.

Prompts that map to real workflows

After installing a skill, paste one of these into Codex, Claude Code, Cursor, or another compatible agent.

The first three workflows are evidence, planning, and CI gates. Write-capable scaffolds and Figma actions remain explicit choices.

Research, stated plainly

The V15 confirmatory audit is a reproducible release study, not a product claim page.

No superiority, speed, or dollar-savings claim is made. The study reports exclusions without imputation and keeps functional, rendered-quality, and resource evidence separate.

Read the conference-style audit PDF, inspect the protocol and receipts, or review the V17 preregistration for the next routing-quality study. The complete InterfaceBench contract and DesignWorkBench v2 readiness report remain separate from release evidence.

Memi InterfaceBench is a 100 target tasks specification with 5 pinned seed tasks; it is not a published performance score. The historical 2.7 candidate record reported 2,187/2,187 tests and 70.57% statements coverage. The greater-than-25% claim remains not verified. Memi DesignWorkBench v2 holds 300 task contracts and requires practitioner calibration before any certification claim. Inspect the benchmark contract and workflow evidence.

How the audit works

1. Inspect — build an evidence graph from source, routes, styles, and local design-system files.

2. Find — report normalized issues with severity, confidence, provenance, and file:line.

3. Correct — let a human or coding agent make a scoped change.

4. Verify — rerun the same command and compare the evidence.

No LLM is used in the deterministic CI enforcement path.

Choose your integration

Design CI

Pin the release commit so every pull request runs the same code:

name: design
on: [pull_request]

permissions:
  contents: read

jobs:
  memi:
    runs-on: ubuntu-latest
    permissions:
      contents: read
      security-events: write
    steps:
      - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
        with:
          fetch-depth: 0
      - uses: memi-design/memi@74fc6ce8c66182b4aa06e1250cb169da8b1fc54c # v2.7.7
        with:
          version: "2.7.7"
          report: true
          upload-sarif: true

The Action adds code-scanning annotations, a step summary, and a memi-design-health artifact. Existing debt can be baselined while new debt fails the gate.

GitHub Action guide · CI recipes · team rollout

Agent and MCP setup

memi agent install codex --project .
memi agent install claude-code --project .
memi agent install cursor --project .
memi agent install grok-build --project .

{
  "mcpServers": {
    "memoire": {
      "command": "memi",
      "args": ["mcp", "start", "--no-figma"]
    }
  }
}

Codex plugin marketplace:

codex plugin marketplace add memi-design/memi --ref main --sparse .agents/plugins --sparse plugins/memoire

Agent stack guide · copy-paste recipes · full skill router

Trust and proof

• Release gates — package, provenance, clean-install, MCP, plugin, binary, and public-surface checks.

• Current release truth — one source for npm, GitHub, Action, Studio, and website versions.

• Reproducible case studies — pinned evidence, abstentions, and paired protocols.

• Audit reports — timestamped findings, evidence gaps, score caps, and owners.

• Dependency trust ledger — direct dependency purpose, dynamic boundaries, and review policy.

• llms.txt — compact machine-readable product map.

Trust defaults: read-only audit; no npm install-time lifecycle scripts; no source upload or covert telemetry; explicit Figma connection; agent-kit --dry-run --json; immutable Action pins; and documented third-party boundaries in NOTICE.

Community

• Ask a question

• Show what Memi found

• Report a bug

• Request a feature

• Contribute

Useful contributions include reproducible audit fixtures, framework adapters, skill improvements, accessible UI cases, motion checks, and real before/after reports.

License

Studio interface references and adapted components include Hermes WebUI and the MIT Warp UI framework boundary around warpui_core and warpui; Warp AGPL application and client code is not copied into Memi.

MIT. See NOTICE for optional adapters and complete third-party attribution.