PlatformAI agent connections · MCP
Supasaito MCP server
Connect Claude, ChatGPT and other agentsto your Supasaito data
Supasaito's MCP server connects Claude, ChatGPT, Cursor and other AI agents to your Supasaito data. Ask questions in the assistant you already use and, with your permission, let agents set up prompts, run scans and take tasks in your campaigns.
Every plan, including Free — read and permitted writes
How the Supasaito MCP server works
From a question in your assistant to work done in Supasaito
Your AI assistant connects to Supasaito through MCP with the permissions you choose. It reads your data, takes the actions you allow and works on campaign tasks — and every step is recorded in Supasaito.
- Your AI assistantClaude, ChatGPT, Cursor or your own agent
- MCP connectionOAuth or a token, with permissions
- SupasaitoYour data, rules and workflow
- Answers and actionsAnalytics, scans, audits, briefs
- Campaign tasksClaimed, done and reviewed
01Connect an AI agent
Connect with OAuth or a personal access token
MCP (Model Context Protocol) is the open standard AI assistants use to work with external tools. Hosted assistants connect to Supasaito with a browser sign-in (OAuth); other clients use a personal access token. Either way, you choose what each connection is allowed to do.
Easiest of all: ask your AI agent to set it up. Client-specific steps and an install guide for agents are in the docs.
02Agents with a name and a face
Give each AI agent its own name, face and colour
An agent token turns an autonomous agent into a teammate you can recognise. Its name, face and colour appear on every task, review and history entry — so you always see which agent did what. A personal token acts as you.
Recognisable everywhere
Tasks, campaign steps, reviews and the work log show the agent’s avatar and name, marked as an agent.
A team of agents
Run a researcher, a copywriter and a publisher side by side, each with its own token and access.
03Data agents can read
Supasaito data available to AI agents
Agents can read visibility, position, sentiment, competitors and sources in every cut, ads in ChatGPT answers, the site audit, Search Console queries, keywords and Google rankings, recommendations and briefs, and your business context. Ask “who gained on us this week, and why?” and get an answer from your own data instead of exporting reports.
04Actions agents can take
What AI agents can change with your permission
With a read + write connection, agents create and update brands, categories, prompts, keywords, competitors and source markup, run scans, audits, keyword research and ranking checks within your plan's quotas, and generate briefs and rewrites. Routine setup and analysis can be delegated while you review the results.
Set up the measurement
Add competitors and aliases, suggest categories and generate prompts for them, add keywords and choose which ones to track in Rankings — setup work you no longer do by hand.
Run scans and audits
Start manual scans, site audits, page deep audits, competitor analysis, keyword research and ranking checks from a conversation when you need fresh data, within your plan's quotas.
Turn recommendations into briefs
Generate a brief or a rewrite, change a recommendation's status and mark it published, without switching tools.
05Campaign tasks for AI agents
AI agents claim and complete campaign tasks
An agent token gives an autonomous agent its own name and avatar. It claims the next task it is allowed to do, reads the brief with the full context and submits the result for review — so you add capacity to campaigns without losing track of who did what.
Personal or agent token
A personal token acts as you. An agent token is a separate actor, so the history shows exactly which agent did what.
Many agents, one set of limits
All agents in a company work within the same plan limits, so parallel work stays predictable.
06Campaign process for agents
Agents follow your campaign process — you stay in control
Connected agents move a campaign through the same steps as your team: idea, preparation, gathering materials, creating and publishing content, collecting results. You decide which steps agents may do and who approves the plan, the materials, the content and the results. An agent cannot skip an approval you have switched on, and you can return work for rework or reopen an earlier step at any time.
Agents take only allowed steps
A step set to “people only” never reaches an agent; “agents only” hands it to them entirely.
Approvals you switch on stay on
Plan, materials, content and results are approved by whoever you choose; review is skipped only when a person sets it so.
Return, reopen, hand back
Send work back with notes, reopen an earlier step, or have an agent release a task so a teammate can pick it up — every change is in the history.
07Permissions and security
Access scopes, roles and revocation for each connection
Every connection is scoped to companies, brands and actions, and calls inherit the same permissions as the dashboard. You can try agents safely: start read-only, widen access later, and cut it off at any time.
Same permissions as the dashboard
Your role — Owner, Admin, Member or Viewer — and the company’s access settings apply to every call, so an agent never sees more than the person who connected it.
Scoped per connection
Choose companies and brands, and whether the agent may change, delete or create campaigns.
Revoke any time
Revoke a token or an app from Settings → History & Access; access ends immediately.
Fair use per company
Each company works within its own plan limits; other companies’ usage is not counted against yours.
08Docs for agents
Tool reference and setup guides
FAQ
Supasaito MCP: common questions
- What is the Supasaito MCP server?
- It lets AI assistants and agents use Supasaito through MCP (Model Context Protocol). A connected agent can read your AI visibility, competitor, source, site and search data and, if you allow it, change settings, run scans and work on campaign tasks.
- Which AI clients can connect?
- Hosted assistants that support remote MCP with OAuth, such as ChatGPT, Claude, Gemini and Perplexity, connect with a browser sign-in. Clients such as Cursor, Cline and your own scripts connect with a personal access token.
- Is MCP included in the Free plan?
- Yes. MCP is available on every plan, including Free, within the plan's quotas and your permissions.
- What can an agent change?
- Only what its connection allows. A read-only connection changes nothing. Read + write lets it create and update brands, categories, prompts, keywords and competitor and source markup; deleting requires a separate permission.
- How do I revoke an agent's access?
- An owner or admin can revoke any token or connected app in Settings → History & Access. Access ends immediately.
Related features
Platform features agents work with