Display/For/Display for Devin
For

Display for Devin

Devin builds it. Display is where your engineering team reviews it. Configure the Display MCP server and Devin can publish its HTML outputs directly to a company-gated URL — without human intervention.

Try Display free →See pricing
Free tier available$49/month flat for TeamsGoogle + Microsoft SSOUnlimited viewers
terminal
$ display publish ./report.html
 
Uploading    47kb
Auth       Google Workspace ✓
 
✓ Published
 
→ https://view.display.dev/p/f474hfd/8f3kx9
 
Viewers    anyone at acme.com
Expires    never
Version    1
 
$ 

Devin produces outputs. Sharing them requires the same broken workflow.

Devin (by Cognition) is an autonomous AI software engineer. It writes code, creates documentation, generates HTML reports — and leaves outputs on disk or in a repository.

When Devin produces an HTML artifact, the sharing problem is the same regardless of which agent generated it:

  • GitHub Pages — requires Enterprise Cloud for private repos. Non-engineers don't have GitHub accounts.
  • Manual deploy — someone still has to run a deploy command. Defeats the point of autonomous generation.
  • Screenshot — kills the interactivity Devin built.
  • Public URL — no company auth on internal outputs.

The right solution: Devin publishes its HTML output autonomously, using an MCP tool, as part of the task.

Agent-native publishing with MCP

Display's MCP server gives Devin a display_publish tool. Add it to Devin's tool configuration:

{
  "mcpServers": {
    "display-dev": {
      "url": "https://mcp.display.dev",
      "apiKey": "your-api-key"
    }
  }
}

Now instruct Devin to publish as part of a task:

"Generate an architecture proposal for the auth migration and publish it to the team."

Devin generates the HTML, calls display_publish(), and returns the company-gated URL in the task summary. No manual publish step. No one needs to be at a terminal.

Or: CLI in the CI step

If Devin's workflow involves a CI pipeline, add Display as a publish step:

- name: Publish Devin output
  run: dsp publish ./devin-output/ --name "task-${{ env.TASK_ID }}"
  env:
    DISPLAY_TOKEN: ${{ secrets.DISPLAY_TOKEN }}

Every Devin task that produces HTML output generates a permanent authenticated URL automatically.

Agentic artifact sharing compared

OptionAutonomous publishCompany authViewer frictionCost
GitHub Pages✅ (with Actions)❌ / $2,100/moHigh (GitHub account)$0 / $2,100/mo
Vercel deploy + SSO✅ (with Actions)✅Medium$320+/mo
Manual screenshot❌—Low (degraded)$0
Display (MCP)✅✅Zero$49/mo flat

Pricing

Free — one-time password auth, 50MB, unlimited viewers
Teams ($49/month) — Google + Microsoft SSO, 25GB, custom domain, unlimited viewers

See all pricing →

Configure Display MCP for Devin →

FAQ

Common questions.

Does Display work as an MCP tool for any agent, not just Devin?+

Yes. Display's MCP server exposes display_publish, display_list, display_get, and display_delete tools. Any MCP-compatible agent — Devin, Claude Code, Cursor, custom agents — can use them.

What types of HTML artifacts does Devin typically produce?+

Devin generates architecture documents, implementation plans, code documentation, test reports, and analysis outputs — often as HTML files. Display publishes all of them.

Can Devin publish multiple artifacts per task?+

Yes. Each display_publish call creates a new artifact. Devin can publish interim artifacts (e.g., a plan before implementation starts) and final artifacts (the completed output).

How do viewers authenticate if they're in different departments?+

Any @yourcompany.com email authenticates — engineering, product, executive, legal. No separate accounts or access provisioning required.

What if Devin produces a multi-file directory?+

display_publish supports directory publishing. The entire output directory is served as a single artifact, with index.html at the root.

Publish your first artifact in 15 seconds.

Free tier. No credit card. One-time password auth on free, Google + Microsoft SSO on Teams ($49/month flat).

Get started free →See pricing