Cleotic

Connected Agents (MCP)

Connected Agents let AI assistants read Cleotic evidence and carry out authorised setup and content workflows through the Model Context Protocol (MCP). Use this when you want an agent client to answer questions about your brands, studies, segments, visibility, share of voice, citations, responses, reports, and Content Studio work without exporting data by hand.

MCP connections start with read access. Explicit write permissions let agents create and update brands, competitors, segments, prompts, and Content Studio work.

What you can connect

Use any agent client that supports remote MCP over Streamable HTTP and bearer-token authentication. OAuth is the preferred connection path. Scoped MCP API keys are available as an alternative route for clients that cannot complete OAuth yet.

The MCP endpoint is shown in Settings -> Connected Agents:

https://mcp.cleotic.ai

Before you start

You need:

  • A Cleotic account with access to the organisation you want the agent to use
  • Accepted Cleotic terms for normal authenticated access
  • An MCP-capable agent client
  • Access to the brands or studies the agent should read

One MCP connection is scoped to the selected organisation. If you work across multiple organisations, create or authorise a separate connection for each one.

Connect an agent client

Use OAuth when your agent client supports it:

  1. In Cleotic, open Settings -> Connected Agents.
  2. Copy the Cleotic MCP endpoint.
  3. In your agent client, add a remote MCP server using that endpoint.
  4. Choose OAuth or browser sign-in when the client asks how to authenticate.
  5. Sign in to Cleotic in the browser and select the organisation you want the agent to read.
  6. Approve the client connection.

After authorisation, the client discovers tools according to its granted scopes and your Cleotic access.

Project IDs

MCP tools refer to each brand or study by its underlying project, using project_id or the project name. list_projects returns each project's kind (brand or study) and accepts a kind filter. See Brands and studies.

Scopes

OAuth connections inherit the signed-in user's current Cleotic organisation and brand access. For MCP API keys, choose the smallest set of scopes the agent needs:

ScopeAccess
mcp.projects.readBrand and study setup, primary brand, competitors, segments, prompts, and selector details
mcp.analytics.readVisibility, share of voice, answer gaps, perception, and source health
mcp.evidence.readResponses, citations, saved reports, and linked source context
mcp.content.readContent Studio settings, opportunities, briefs, generation steps, skills, and visibility goals
mcp.account.readPlan, usage, capacity, and capability information
mcp.docs.readPublic Cleotic docs. Docs tools are available to authenticated MCP sessions.
mcp.projects.writeCreate and update brands, the primary brand, competitors, segments, and prompts
mcp.content.writeCreate and update briefs, brand skills, visibility goals, opportunities, and settings
mcp.content.generateStart AI content generation for briefs

Write permissions also require the matching read permission. The key form selects it automatically. Generating content uses your plan allowances and model budget. Brand setup can also start scheduled collection or initial prompt runs.

OAuth write access requires the client to request and receive the relevant write scopes. If the client cannot obtain them, create a new MCP key with the permissions you need. Existing keys keep their original grants.

If an MCP API key scope is missing, Cleotic either hides the related tool from discovery or returns a structured missing-scope error for that tool call.

Available tools

The exact tool list depends on the signed-in user's access for OAuth connections, or the selected key scopes for MCP API keys.

Documentation

  • search_docs
  • get_doc

Brands, studies, and setup

  • list_projects
  • get_project_overview (setup includes primary_brand and competitors)
  • list_segments
  • list_prompts

Analytics

  • get_portfolio_summary
  • get_visibility
  • get_share_of_voice
  • get_answer_gaps
  • get_perception_summary
  • get_source_health

Evidence

  • list_responses
  • get_response
  • list_citations
  • get_citation
  • list_reports
  • get_report

Content Studio

  • get_content_studio_settings
  • list_content_opportunities
  • get_content_opportunity
  • list_content_briefs
  • get_content_brief
  • get_content_brief_run_steps
  • list_content_skills
  • list_visibility_goals

Visibility goals are readable through MCP where present; the app has no separate goal-management screen.

Account

  • get_plan_usage

Cleotic also exposes guided prompts such as a brand health brief and an evidence drill-down when the connection has the required scopes.

Write workflows

With mcp.projects.write and mcp.projects.read:

  • create_project, update_project
  • set_primary_brand
  • create_competitor, update_competitor
  • create_segment, update_segment
  • create_prompt, update_prompt, replace_prompt

With mcp.content.write and mcp.content.read:

  • update_content_studio_settings
  • create_content_brief, update_content_brief
  • update_content_opportunity
  • create_content_skill, update_content_skill
  • create_visibility_goal, update_visibility_goal

With mcp.content.generate and mcp.content.read:

  • generate_content_brief

Write calls require an idempotency_key. Agents should reuse the same key and arguments when retrying an action, and use a new key for a different action. Use get_mcp_operation to follow an operation's progress.

Suggested first prompts

After connecting, try:

  • "List my Cleotic brands and summarise which one needs attention first."
  • "Give me a health brief for the brand named Acme."
  • "For Acme, compare visibility and share of voice over the last 30 days."
  • "Find answer gaps where competitors appear and my brand does not."
  • "Pull supporting response and citation IDs for the strongest recommendation."

Ask the agent to cite Cleotic response, citation, or report IDs when it makes evidence-backed claims.

Manage and revoke keys

Open Settings -> Connected Agents to review active MCP keys. Revoke a key when:

  • A teammate leaves the organisation
  • A laptop, server, or agent client is retired
  • A secret may have been pasted somewhere unsafe
  • The client no longer needs Cleotic access

Revoking a key stops future MCP requests that use it. OAuth connections follow the connected user's current Cleotic permissions.

Troubleshooting

Check that the endpoint is https://mcp.cleotic.ai and that the client is configured for remote MCP over Streamable HTTP.

Select Streamable HTTP in the client's transport settings. If it tries to use a different transport, use the client's remote HTTP setup instructions.