agentzia
Menu

Agentzia MCP.
Let your agent find the fix.

Connect Codex, Claude Code, or Cursor to Agentzia. Your coding agent gets the evidence to improve your page and win more AI recommendations, then retests the changes.

Client installation

Connect the AI client you already use.

Choose Codex, Claude Code, Cursor, Grok Bot, or any OAuth-capable MCP client. Browser sign-in keeps credentials and payment details out of the agent's context.

  1. 01
    Add the server

    Copy the commands above into your terminal. They register Agentzia for this client and begin secure sign-in.

  2. 02
    Approve sign-in

    Your browser opens Agentzia. Sign in and approve the connection; your MCP client keeps the resulting OAuth session.

  3. 03
    Start a fresh session

    Restart any session that was already open so the client loads Agentzia's seven tools from the beginning.

  4. 04
    Run your first test

    Ask your agent: “Use Agentzia to analyze this landing page, explain the report, and propose the highest-priority improvements.”

Agentzia MCPConnect your coding agent
OAuth ready
Secure MCP endpointhttps://www.agentzia.dev/mcpStreamable HTTP
CodexTerminal
codex mcp add agentzia --url https://www.agentzia.dev/mcp
codex mcp login agentzia

Add Agentzia, then sign in securely in your browser. No token to copy.

Connect with Codex No API key or private token required.

Recommended loop

Ship a better page. Check if AI chooses it.

Read the complete agentic workflow
  1. 01

    Check list_runs for reusable research, then get_credits and choose the AI assistants to test.

  2. 02

    Call analyze_url with the public landing-page URL, clients and a UUID requestId. Reuse the same ID and arguments if the response is lost.

  3. 03

    Poll get_run with section: "status" at the returned pollAfterMs interval until terminal is true. Check reportQuality and missing participants before interpreting results.

  4. 04

    Read summary and actions first. Follow nextOffset for more items. Follow an action’s evidenceCalls, or read verdicts with a specific scenarioId and detail: "full" for exact responses, citations and traces.

  5. 05

    Verify the evidence, make the revision, and retest the same URL within the user-approved credit budget. Read comparison caveats before claiming an improvement.

Available tools · 7

01list_runs

Find existing tests before spending credits on new research.

02get_credits

Check channel credits, plan and advisory free-capacity availability.

03analyze_url

Test a public product page across selected AI channels, at one credit per channel.

04analyze_content

Evaluate supplied public-safe marketing copy when a public URL is unavailable.

05get_run

Poll progress or read a focused report section and its supporting evidence.

06retry_run

Resume a failed or incomplete test using saved research without another channel credit.

07create_credit_checkout

Create a browser payment link for the human when more credits are needed.

From finding to fix

Clear for your team. Ready for your agent.

BrowserA visual report for your team

Search results, recommendations, trust concerns, competitors, page reach, methodology, prioritized changes, and every completed response from your selected channels.

Open the samples
MCPStructured evidence for your agent

Compact summaries, paginated actions, direct evidence calls and completed page checks help your agent make a supported edit and retest. Unknown visits remain unknown; unresolved findings are kept separate from actions.

Open raw JSON

Endpoint

https://www.agentzia.dev/mcp

Auth Browser-based OAuth 2.1 with PKCE.

Output Structured summary, actions, per-client verdicts, methodology, complete responses, citations, traces, and a complete export.

Evidence Actions include evidenceCalls for exact supporting records and referenceChecks for completed public page reads. HTTP, canonical and robots metadata are included when returned. These checks do not prove indexing or a participant visit.

Interpretation Unmeasured visits are null, not zero. Trust concerns, blockers and unknown judgments are distinct. Product-fit and routine pricing caveats alone do not establish distrust. Exact searches and visits depend on what each provider exposes; citations alone do not verify a visit.

Recovery Check retryable before retry_run. It reuses saved work without another credit, including incomplete reports. For a new test after edits, use a new requestId. Check reportCurrent while a retry is running.

Billing Every account starts with complimentary credits. Pro, Ultra, and Scale include monthly credits; optional top-ups and all payment steps stay in the browser.

Data Public marketing content only. Read retention, deletion, and external processing details.

Context cost Default reads return summaries; actions and verdicts return three items per page. Use insights for coverage, competitors and unresolved findings, and search for completed reference checks. full remains available for export and backward compatibility.

Ready to connect

Find what holds your product back. Let your agent help fix it.

Connect my coding agent