# TheDesignAgent

> A design agent in your coding agent's loop: briefs each screen, reviews UX and visual quality.

- name: ai.thedesignagent/mcp
- version: 0.5.3
- connection class: R2 (oauth_consent_once)
- trust: 83/100 flags: no-repository
- quality: 92/100 (strong)
- owner: not claimed
- descriptor: https://api.protogrid.dev/v1/servers/ai.thedesignagent%2Fmcp
- tools: https://api.protogrid.dev/v1/servers/ai.thedesignagent%2Fmcp/tools

**An agent can connect after a one-time OAuth consent by a human.** Remote endpoint behind OAuth 2. Tokens stay with the agent; the registry only exposes the authorization metadata.

## Connection (mcpServers)

```json
{
  "mcpServers": {
    "thedesignagent": {
      "type": "http",
      "url": "https://www.thedesignagent.ai/mcp",
      "headers": {
        "Authorization": "Bearer ${THEDESIGNAGENT_TOKEN}"
      }
    }
  }
}
```

Placeholders to fill: `${THEDESIGNAGENT_TOKEN}`.

## Trust

- hygiene: 80
- liveness: 100
- freshness: 100
- provenance: 55
- drivers: +dns-namespace +website +reachable +uptime-100% +updated-14d-ago -no-repository
- Derived from observable signals (official registry feed and our own credential-free probes); no code audit performed.

## Quality 92/100 (strong)

### protocol: n/a (weight 25)

- n/a `protocol.modern`, `protocol.stateless`, `protocol.transport`, `protocol.list_ttl`: no remote answered a protocol handshake (not probed yet, unreachable, or local-only)

### authorization: 83 (weight 15)

- pass `auth.prm`: Protected Resource Metadata (RFC 9728) published
- pass `auth.as_metadata`: authorization server metadata reachable
- warn `auth.cimd`: only Dynamic Client Registration, deprecated in 2026-07-28 in favour of CIMD
- n/a `auth.secret_in_url`: no templated URL

### tool hygiene: 93 (weight 30)

- pass `tools.descriptions`: every tool has a description
- warn `tools.description_length`: 1 tools have descriptions over 1,024 characters, which crowds the context
- pass `tools.schemas`: every tool has an object input schema
- pass `tools.annotations`: 3 of 3 tools declare readOnlyHint or destructiveHint
- pass `tools.directory_hints`: every tool declares readOnlyHint, destructiveHint, idempotentHint and openWorldHint
- pass `tools.token_cost`: about 1,242 tokens to load every tool
- pass `tools.api_dump`: tools are not a one-to-one API dump

### stability: 100 (weight 15)

- pass `stability.changes`: 3 tool changes in 30 days
- pass `stability.rug_pull`: no tool changed its meaning under the same name

### dependencies: n/a (weight 15)

- n/a `deps.known_vulns`, `deps.mcp_sdk_version`, `deps.resolvable`: deps.dev does not know @thedesignagent/mcp@0.5.3

- tools: 3, about 1,242 tokens to load them all
- tool set hash: 2bc46223505050c0a7fedfbc65ec1c08
- tools last changed: 2026-10-04T20:26:40.418Z
- badge: [![protogrid quality](https://protogrid.dev/badge/ai.thedesignagent/mcp.svg)](https://protogrid.dev/servers/ai.thedesignagent/mcp)
- Checks run on what our credential-free, read-only probes observe; tools are never called and no code audit is performed.

### Tool changes

- 2026-10-04 `Visual` changed (title, outputSchema, annotations)
- 2026-10-04 `Ux` changed (title, inputSchema, outputSchema, annotations)
- 2026-10-04 `Discover` changed (title, outputSchema, annotations)

Full history: https://api.protogrid.dev/v1/servers/ai.thedesignagent%2Fmcp/changes

## Dependencies

- npm @thedesignagent/mcp@0.5.3: unknown

Dependency graphs from deps.dev; advisories from OSV.dev, including the GitHub Advisory Database and the PyPI Advisory Database; all CC BY 4.0.

## Remotes

- https://www.thedesignagent.ai/mcp (streamable-http, auth oauth2, reachable true, uptime30d 1)

## Packages

- npm @thedesignagent/mcp@0.5.3 secrets: THEDESIGNAGENT_API_KEY

## Tools (3)

- `Discover`: Get a build brief calibrated to your project, persona, and the specific screen you're building.

Call this BEFORE building any screen. Returns:
- The active Job-to-be-Done for the task
- Relevant heuristics, cognitive guidance, and patterns from the corpus
- A focused build brief (~600 tokens) to inject into your build context

CALL LIFECYCLE:
- First call for a new project (no project_id): returns _meta.tda_status='setup_required' with an extraction task. Scan the repo's schema/types/migrations (or, with no repo, ask the user for a short product description as the task explains) and resubmit with project_model populated.
- Subsequent calls: send ONLY task + project_id. project_model is cached server-side from the first call; resending it wastes ~2–4k tokens per call and re-runs validation. Resend project_model ONLY when a response returns _meta.tda_status='setup_required' or 'model_stale'.

Use project_id from .thedesignagent or memory across calls. Pass repo_hash (SHA256 of git remote origin URL) only as a fallback if no project_id is known yet.
- `Ux`: Evaluate a built UI artifact for JTBD fit, heuristic compliance, cognitive load, and pattern conformance.

Call this AFTER building a screen. TheDesignAgent scores:
- Heuristics: violations against Nielsen's heuristics + project-specific anti-patterns
- Cognitive: load profile against the stored persona
- Patterns: conformance to established project pattern decisions
- JTBD fit: does the artifact serve the resolved job?

Returns ux_score (0–10) + findings + prioritised recommendations.

Stores new anti-patterns to project context — future builds benefit automatically.
- `Visual`: Evaluate a built UI artifact for visual design quality — brand compliance, visual cognitive load, and aesthetic quality.

Pass render_url to review a real screenshot. This hosted server screenshots public pages and preview deploys (pass vercel_bypass_token for a protected Vercel preview). It cannot reach localhost or private networks: for those, it returns instructions for the local MCP server. With no render_url, the review runs from the artifact code alone.

If the page redirects to login, no review runs and nothing is charged.