Theysaid

Launch AI usability tests on live sites, Figma prototypes and designs, then read results from Claude.

Opens Agent Studio, where connecting is one click. The connector URL below works in any MCP client.

Paste it into any MCP client. Setup docs

What the Theysaid connector does

The TheySaid MCP server lets Claude launch AI usability tests against live URLs, Figma prototypes or designs built in tools like v0, Bolt and Lovable. AI testers navigate the product and report friction points in plain language. From the same conversation you can check test status, explore themes and compare projects. Sign-in is OAuth, and it can create your account.

Verified connector

Listed by Anthropic as a partner connector in its Connectors Directory.

Connection checked by Agentman on .

Anthropic states this reflects the level of review a connector received, not a security audit.

Theysaid tools (33)

  • channel_breakdown
  • check_design_snapshot
  • company_data
  • contact_detail
  • contact_list
  • cross_project_comparison
  • flag_result
  • generate_chart
  • get_question
  • get_test_results
  • get_test_status
  • insight_video_metadata
  • launch_usability_test
  • list_projects
  • org_documents
  • ping
  • prepare_design_snapshot
  • project_estimated_duration
  • project_overview
  • project_stats
  • propose_usability_test
  • question_breakdown
  • recent_responses
  • relaunch_usability_test
  • response_detail
  • responses_search
  • revise_usability_test
  • search_ai_tester_voices
  • similar_projects
  • stats_trend
  • theme_detail
  • upload_design_snapshot
  • wait_for_results

Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.

Limits

  • Documentation not publicly readable to us. The directory's documentation link is a ClickUp doc that renders only in a browser running its scripts; our fetches returned an empty loading page. This page is built from the directory listing and our live probe.
  • Cost of a test unconfirmed. We could not establish whether launches spend credits or quota.
  • Tool behaviour inferred from names. Without the tool list or docs, effects above are what the names indicate, not what the server states.
  • Prompts unverified. The directory lists three prompts; we could not list them.
  • We called no tool. Our probe was an anonymous initialize and the RFC 9728 metadata.

Sources

Use it in an agent

Put Theysaid to work.

Connect Theysaid once and your agents call these tools on their own: on a schedule, in a workflow, with nobody at the keyboard.

Call (650) 285-1019, Agentman.Our AI receptionist answers.