For the complete documentation index, see llms.txt. This page is also available as markdown.
MCP access, permissions & audit
Learn who can connect an AI client to Convia, which workspaces it reaches, how plans, limits, tokens and OAuth work, and how to audit or cut off access.
On this page
- Which roles can use MCP Connections
- MCP plans: MCP Connections and Advanced MCP
- MCP limits per workspace
- MCP workspace scoping
- What each request is checked against
- Connected apps in a workspace
- Apps you trusted
- MCP token lifecycle
- Removing a member stops their MCP access
- OAuth for client developers
- MCP connection errors
- The MCP Activity log
- What MCP never does
- Related
Convia Pro checks every MCP request against the person behind the credential, their role in each workspace the connection reaches, and each workspace's plan, before any tool runs. Only workspace Owners and Managers can connect an AI client, every tool call is recorded on the Activity tab, and removing a member, revoking a token, disconnecting an app or changing plan takes effect on the next request.
Which roles can use MCP Connections
MCP Connections is limited to the people who manage a workspace.
| Role | Can create tokens | Can connect a client | Can see and revoke tokens and Activity |
|---|---|---|---|
| Owner | Yes | Yes | Yes |
| Manager | Yes | Yes | Yes |
| Contributor | No | No | No |
| Viewer | No | No | No |
The role is checked on every request, not only when a connection is made. If an Owner or Manager is changed to Contributor or Viewer, their tokens and connectors stop reaching that workspace on the next request. Roles are described in Workspaces & teams.
A token always acts as the person who created it. Owners and Managers can see every token in the workspace and revoke any of them, but they can only create tokens for themselves.
A partner workspace cannot be connected until the client claims it, and a partner's seat in a client's workspace is never a Manager seat, so a partner cannot connect an assistant to it.
MCP plans: MCP Connections and Advanced MCP
MCP Connections comes with Grow and above. Without it, the MCP Connections page shows a locked card, and every request from an existing token or connector is refused with feature_disabled. The plan is checked on every request, so if a plan change removes MCP Connections, all of the workspace's connections stop at once.
Some tools need more on the plan, checked each time they are called:
- Advanced MCP, on Scale and above: drafting or revising up to 100 posts in one step, and changing a video already on YouTube.
- Product features, such as Advanced Analytics for the analytics tools, Magic Campaign for the campaign plan tools, and Ideas and Scripts for their tools. An assistant gets the same features the app gives that plan.
A client is offered only the tools the plan includes. The MCP Connections page lists every tool and marks the others Not on your plan (or Advanced MCP) with a link to the plans. After an upgrade, reconnect the assistant so it sees the new tools. The full table is in the MCP tool reference, and plans are on Plans & billing.
MCP limits per workspace
Each workspace has a limit on how much connected assistants can do, shared by every connection that reaches it.
| Plan has | Requests per hour | Changes per day |
|---|---|---|
| MCP Connections | 300 | 100 |
| Advanced MCP | 1,000 | 500 |
The hour and the day roll, so once a limit is reached the assistant is told to wait a few minutes and try again (rate_limited). Reading still works when only the change limit is reached. A batch counts each post it changes, and a preview of a YouTube video change counts nothing until it is applied. Separately, one credential can send at most 60 requests a minute.
MCP workspace scoping
One address, https://mcp.conviapro.com, serves every workspace. What a connection reaches depends on how it signs in.
- A connector (OAuth) reaches the workspaces ticked on the Connect screen when it was approved: those the person owns or manages on a plan with MCP Connections, up to 25. The choice narrows what the person can reach and never widens it. Every request re-checks membership, role and plan in each of those workspaces, so a workspace the person loses drops out and the rest keep working.
- Several workspaces. Every tool call must name one, and every answer says which workspace it came from. A call naming none, or one the connection was not given, is refused rather than guessed.
- A token belongs to the workspace it was created in. Used with an address naming another workspace, it is refused with
workspace_mismatch. - Older connectors with an address ending in a workspace id keep working. One approved before workspaces could be chosen reaches any workspace its person owns or manages through such an address; once a choice has been saved for it, it is held to that choice.
- Every tool reads and writes only the workspace it runs in. An id from another workspace gets the same answer as an id that does not exist.
To change which workspaces a connector reaches, remove it in the assistant and add it again.
What each request is checked against
The Convia MCP server runs these checks on every request. The first that fails ends the request, and no tool runs.
- The credential is valid. A token must exist and not be revoked or expired. An OAuth access token must be valid, issued to an MCP client, and, for an app Convia does not recognize, trusted by the person.
- The workspace is one the connection may reach: the token's own, or one ticked for the connector.
- The person behind it is still a member of the workspace.
- They are still an Owner or Manager.
- The workspace's plan includes MCP Connections.
- The workspace is within its request and change limits, checked when a tool is called.
Because these run every time, removing a member, changing a role, revoking a token, disconnecting an app or changing plan takes effect on the very next request.
Connected apps in a workspace
The Connected apps tab on Integrations → MCP Connections lists the apps connected to the workspace you're in, who connected each one, when it was last used and when it was connected. Owners and Managers see everyone's; anyone else sees their own. It includes apps connected before workspaces could be chosen.
Disconnect from this workspace, in a row's menu, stops that app reaching the workspace at its next request, without touching the other workspaces it reaches. It cannot be undone: to give the workspace back, remove the connector in the app and add it again. Disconnecting ends the connection, not the person: to stop a person using MCP, change their role. Steps are in Connect an AI assistant (MCP).
Apps you trusted
The Connect screen shows Sends you back to, the address the approval returns to, and judges it before you can approve:
| Address | What happens |
|---|---|
| A known AI assistant (Claude, ChatGPT or Grok) | Approve as normal |
| An app on this computer, such as Claude Code | Approve only if you just started that connection yourself |
| Any other secure web address | A warning, and I trust followed by the address must be ticked before Approve works |
| An unsafe address | Cannot be approved |
The app's name on the screen is chosen by whoever set it up and proves nothing; the address is what counts. An app you trust connects for you only. The Apps you trusted tab lists each one with the address it sends you back to. Stop trusting cuts the app off at its next request, in every workspace. To use it again, connect it again and tick I trust.
MCP token lifecycle
Workspace tokens are for clients that send a fixed header, such as Claude Code. Create and revoke them in Integrations → MCP Connections; the steps are in Connect an AI assistant (MCP).
| Stage | What happens |
|---|---|
| Format | mcp_live_ followed by 64 hexadecimal characters |
| Creation | Shown once, in the Your MCP token dialog. The dialog cannot be closed until you tick I have saved this token in a safe place |
| Storage | Convia keeps only a SHA-256 hash of the token, so it cannot show it again or recover it. The list shows the first 12 characters to help you tell tokens apart |
| Expiry | Never, 30 days or 90 days, chosen at creation. An expired token is refused with token_expired and marked Expired |
| Last used | Updated as the token is used, at most about once a minute |
| Revocation | Revoke on the token's row. Immediate and permanent. The token is kept, marked Revoked, so its past activity stays attributed |
| Number of tokens | A workspace can have several tokens. Give each client its own, so you can revoke one without breaking the others |
Treat a token like a password. Anyone holding it can act in the workspace as you, with your role, until it is revoked or expires.
Removing a member stops their MCP access
Removing a person from the workspace stops every token and connector they use there, on the next request (not_a_member). Their tokens are not revoked automatically, though: they stay in the Tokens list.
OAuth for client developers
Connectors sign in with OAuth 2.1 instead of a token. A client developer needs the following.
-
Discovery. A request with no credential gets HTTP
401with a challenge pointing at the protected resource metadata:textWWW-Authenticate: Bearer realm="mcp", resource_metadata="https://mcp.conviapro.com/.well-known/oauth-protected-resource"A request with a credential that is not accepted gets the same header with
error="invalid_token"added. The header is exposed to browser clients through CORS. -
Protected resource metadata.
GET https://mcp.conviapro.com/.well-known/oauth-protected-resourceis public and returns:json{ "resource": "https://mcp.conviapro.com", "authorization_servers": ["<authorization server issuer>"], "bearer_methods_supported": ["header"], "scopes_supported": ["openid", "email", "profile", "offline_access"] }resourceis the MCP address the client called. Fetch the authorization server's own metadata from the issuer inauthorization_serversusing standard OAuth discovery. -
Client registration. Dynamic client registration is available: register through the
registration_endpointin the authorization server's metadata. Register as a public client and use PKCE withS256. A client whose redirect addresses are not a known AI assistant or an app on the person's own computer works only for a person who ticks I trust for it on the Connect screen. -
Consent. The person signs in to Convia and sees Connect followed by the app's name, Signed in as, Sends you back to and Workspaces it can use. They tick workspaces and choose Approve or Deny.
-
Scopes. Only the standard scopes above exist; there are no custom scopes. Request
offline_accessto receive a refresh token. -
Calls. Send the access token as
Authorization: Bearer <access token>on every request. If the person ticked several workspaces, every tool takes aworkspaceargument; see MCP protocol, connections & errors.
Convia refuses an ordinary Convia sign-in session token sent as a bearer (oauth_token_required). A 403 never carries a challenge: the credential was fine but the person or workspace is not allowed, so signing in again will not help.
MCP connection errors
Connection errors happen before any tool runs and are returned as HTTP errors with a JSON body such as {"error":"token_revoked"}. Tool-level errors, including rate_limited, are listed in MCP protocol, connections & errors.
| HTTP status | Error | Meaning and fix |
|---|---|---|
| 400 | invalid_workspace_id |
An older address does not end in a valid workspace ID. Use https://mcp.conviapro.com |
| 400 | unsupported_protocol_version |
The MCP-Protocol-Version header is not one Convia supports. The response lists the supported versions |
| 401 | missing_bearer |
No Authorization: Bearer header |
| 401 | invalid_token |
The token or access token is not recognized |
| 401 | token_revoked |
The token was revoked. Create a new one |
| 401 | token_expired |
The token passed its expiry date. Create a new one |
| 401 | workspace_mismatch |
The token belongs to a different workspace than the address names |
| 401 | oauth_token_required |
The bearer is a normal sign-in session, not an OAuth access token |
| 401 | client_not_allowed |
The access token was issued to an OAuth client Convia does not accept |
| 401 | client_not_trusted |
The app is not one Convia recognizes and the person has not trusted it, or it sends people to an unsafe address. Reconnect and tick I trust, if you trust it |
| 403 | workspace_not_granted |
The connection was not given this workspace, or has none. Remove the connector and add it again |
| 403 | not_a_member |
The person is no longer a member of the workspace |
| 403 | insufficient_role |
The person is not an Owner or Manager |
| 403 | feature_disabled |
The workspace's plan does not include MCP Connections |
| 405 | method_not_allowed |
Only POST is accepted, apart from the metadata document |
| 429 | rate_limited |
More than 60 requests in a minute from one credential. Wait for Retry-After (60 seconds) |
| 503 | client_trust_unavailable, grant_lookup_failed, workspace_lookup_failed |
Convia could not check the connection just now. Retry shortly |
The MCP Activity log
Every tool call made through MCP is recorded in the workspace it ran in, and Owners and Managers can review it.
- Open Integrations → MCP Connections.
- Click the Activity tab.
Each row shows When, Token (the token's name), Tool, Result (Success, or the error code) and Latency. Rows are newest first, 50 to a page.
| Recorded | Not recorded |
|---|---|
Every tools/call, successful or failed, including calls refused by the workspace's limits |
Requests refused before a tool runs: credential, membership, role or plan failures, the per-credential flood limit and malformed requests |
| A short summary of the arguments (counts, lengths, flags), never post content | initialize, ping and tools/list |
The first 200 characters of the reason given to remove_posts |
The argument summary and the reason are stored with each entry but are not shown on the Activity tab. Calls through a connector are not tied to a token, so the Token column shows (deleted token) for them. The log is append-only: nobody in the workspace can edit or delete entries.
What MCP never does
Whatever the credential, the Convia MCP server cannot:
- Publish a post, schedule one to go out, or approve one. New posts are always written as drafts. An assistant can send drafts to the approval queue, where a person decides. Editing a post a person already scheduled leaves it scheduled. See Approval & review.
- Edit, unschedule or trash a published or approved post. The one exception is changing the title, description, chapters or tags of a video Convia uploaded to YouTube, with Advanced MCP, after a preview the person agrees to.
- Change the prompt Convia uses now. It can only stage an alternative or a draft for a person to make live.
- Set TikTok privacy or branded-content options, or YouTube's Made for Kids, privacy status or publish time.
- Add anything but images and MP4 or QuickTime videos to the Media library. Videos come only from a local file upload, not from a link.
- Reach a workspace the connection was not given, or one where its person is not an Owner or Manager.
Using MCP spends no AI credits: your assistant writes the copy on your own subscription with that assistant.
Related
- Connect an AI assistant (MCP): create a token or a connector step by step.
- MCP tool reference: tool-level errors, plan requirements and side effects.
- Workspaces & teams: roles and how to change them.
- Plans & billing: which plans include MCP Connections and Advanced Analytics.