roboto.ai.mcp

Submodules

• roboto.ai.mcp.operations
• roboto.ai.mcp.record

Package Contents

class roboto.ai.mcp.AdminRegisterMcpServerRequest(/, **data)

Bases: pydantic.BaseModel

Admin request to register an MCP server. Supports both DCR and manual modes.

Admins can scope the server to specific orgs via org_ids, or leave it empty for global visibility.

Parameters:

data (Any)

allowed_tools: dict[str, bool] | None = None

map of tool name to enabled flag. None = configure later (auto-populates on first discovery).

Type:

Tool allowlist

authorization_endpoint: str | None = None

OAuth2 authorization endpoint URL.

client_id: str | None = None

OAuth2 client ID from the provider’s developer settings.

client_secret: str | None = None

OAuth2 client secret from the provider’s developer settings.

display_name: str = None

User-friendly display name (e.g., ‘GitHub’). Alphanumerics, hyphens, and underscores only.

extra_auth_params: dict[str, str] = None

Extra query parameters appended to the OAuth authorization URL. For example, Atlassian requires {“audience”: “api.atlassian.com”, “prompt”: “consent”}.

oauth_issuer_url: str | None = None

//github.com’).

Type:

OAuth2 authorization server base URL (e.g., ‘https

org_ids: list[str] = None

Org IDs this server is scoped to. Empty = global (visible to all orgs).

registration_mode: RegistrationMode

revocation_endpoint: str | None = None

OAuth2 token revocation endpoint URL, if supported.

scopes: list[str] = None

OAuth2 scopes to request during authorization.

server_url: str

MCP server Streamable HTTP endpoint URL.

token_endpoint: str | None = None

OAuth2 token endpoint URL.

class roboto.ai.mcp.CopyMcpServerAllowedToolsRequest(/, **data)

Bases: pydantic.BaseModel

Admin request to copy the tool allowlist from another server.

Parameters:

data (Any)

source_server_id: str

Server ID to copy the allowed_tools dict from.

class roboto.ai.mcp.McpServerOrgContextRecord(/, **data)

Bases: pydantic.BaseModel

Per-org context for an MCP server.

Org admins set this plaintext to provide the AI with org-specific instructions about how the server should be used (e.g., which repos, project names, conventions).

Parameters:

data (Any)

context: str

Plaintext context injected into the AI system prompt.

created: datetime.datetime

created_by: str

modified: datetime.datetime

modified_by: str

org_id: str

Org that owns this context.

server_id: str

Server this context applies to.

class roboto.ai.mcp.McpServerRecord(/, **data)

Bases: pydantic.BaseModel

A registered external MCP server. May be global or scoped to specific orgs.

Parameters:

data (Any)

allowed_tools: dict[str, bool] | None = None

Map of tool name to enabled flag. None = not yet configured (auto-populates on first discovery). Keys are all known tools from the server; True = allowed, False = blocked.

client_id: str

OAuth2 client ID.

created: datetime.datetime

Timestamp when this server was registered.

created_by: str

User who first registered this server.

display_name: str

User-friendly display name (e.g., ‘GitHub’, ‘GitLab’).

extra_auth_params: dict[str, str] = None

Extra query parameters appended to the OAuth authorization URL (e.g. audience for Atlassian).

modified: datetime.datetime

Timestamp when this server record was last modified.

modified_by: str

User who last modified this server record.

oauth_issuer_url: str

OAuth2 authorization server URL for this MCP server.

org_ids: list[str] = None

Org IDs this server is scoped to. Empty = global (visible to all orgs).

registration_mode: str = 'dcr'

‘dcr’ or ‘manual’.

Type:

How this server was registered

scopes: list[str] = None

OAuth2 scopes requested during authorization.

server_id: str

Unique identifier for this server registration.

server_url: str

MCP server Streamable HTTP endpoint URL.

status: McpServerStatus

Current status of the server registration.

class roboto.ai.mcp.McpServerStatus

Bases: roboto.compat.StrEnum

Status of a registered MCP server.

ACTIVE = 'active'

Server is registered and ready for use.

ERROR = 'error'

Server registration encountered an error.

PENDING_SETUP = 'pending_setup'

Server registration is in progress (e.g., awaiting Dynamic Client Registration).

class roboto.ai.mcp.McpTokenRecord(/, **data)

Bases: pydantic.BaseModel

Per-user, per-org OAuth token metadata for an MCP server.

Token values (access_token, refresh_token) are never exposed — this record contains only status metadata for display in the UI.

Parameters:

data (Any)

created: datetime.datetime

Timestamp when this token was first created.

expires_at: datetime.datetime | None = None

When the access token expires, if known.

modified: datetime.datetime

Timestamp when this token was last refreshed or modified.

org_id: str

Org context this token is scoped to.

scopes: list[str] = None

Scopes granted by this token.

server_id: str

Server this token is for.

token_status: McpTokenStatus

Current status of the token.

user_id: str

User who authorized this token.

class roboto.ai.mcp.McpTokenStatus

Bases: roboto.compat.StrEnum

Status of a user’s OAuth token for an MCP server.

EXPIRED = 'expired'

Access token has expired. May be refreshable via refresh token.

REFRESH_FAILED = 'refresh_failed'

Refresh token failed. User must re-authorize.

VALID = 'valid'

Token is valid and can be used for requests.

class roboto.ai.mcp.McpToolInfo(/, **data)

Bases: pydantic.BaseModel

Describes a tool discovered from a remote MCP server.

Parameters:

data (Any)

description: str

Human-readable description of the tool.

input_schema: dict[str, Any]

JSON Schema describing the tool’s input parameters.

name: str

Tool name as reported by the MCP server.

server_id: str

Server this tool belongs to.

class roboto.ai.mcp.OAuthCallbackRequest(/, **data)

Bases: pydantic.BaseModel

Request to complete an OAuth flow after the user has authorized.

Parameters:

data (Any)

code: str

Authorization code from the OAuth provider.

state: str

CSRF state token from the original flow start.

class roboto.ai.mcp.RegisterMcpServerRequest(/, **data)

Bases: pydantic.BaseModel

Request to register an MCP server. Used by org admins.

In DCR mode, OAuth metadata is auto-discovered and client credentials are obtained via Dynamic Client Registration.

In Manual mode, the caller provides all OAuth credentials upfront.

Parameters:

data (Any)

allowed_tools: dict[str, bool] | None = None

map of tool name to enabled flag. None = configure later (auto-populates on first discovery).

Type:

Tool allowlist

authorization_endpoint: str | None = None

OAuth2 authorization endpoint URL.

client_id: str | None = None

OAuth2 client ID from the provider’s developer settings.

client_secret: str | None = None

OAuth2 client secret from the provider’s developer settings.

display_name: str = None

User-friendly display name (e.g., ‘GitHub’). Alphanumerics, hyphens, and underscores only.

extra_auth_params: dict[str, str] = None

Extra query parameters appended to the OAuth authorization URL. For example, Atlassian requires {“audience”: “api.atlassian.com”, “prompt”: “consent”}.

oauth_issuer_url: str | None = None

OAuth2 authorization server URL. Required for DCR if different from server_url. Required for Manual mode.

registration_mode: RegistrationMode

revocation_endpoint: str | None = None

OAuth2 token revocation endpoint URL, if supported.

scopes: list[str] = None

OAuth2 scopes to request during authorization.

server_url: str

MCP server Streamable HTTP endpoint URL.

token_endpoint: str | None = None

OAuth2 token endpoint URL.

class roboto.ai.mcp.RegistrationMode

Bases: str, enum.Enum

str(object=’’) -> str str(bytes_or_buffer[, encoding[, errors]]) -> str

Create a new string object from the given object. If encoding or errors is specified, then the object must expose a data buffer that will be decoded using the given encoding and error handler. Otherwise, returns the result of object.__str__() (if defined) or repr(object). encoding defaults to ‘utf-8’. errors defaults to ‘strict’.

DCR = 'dcr'

MANUAL = 'manual'

class roboto.ai.mcp.SetMcpServerOrgContextRequest(/, **data)

Bases: pydantic.BaseModel

Request to set per-org context for an MCP server.

The context is plaintext that gets injected into the AI system prompt so the AI knows how the org uses this server. Bounded to keep prompt size predictable and to limit the blast radius of injection from org-controlled text. Set to empty string to clear.

Parameters:

data (Any)

context: str = None

Plaintext context (max 4000 chars). Empty string clears it.

class roboto.ai.mcp.StartOAuthFlowResponse(/, **data)

Bases: pydantic.BaseModel

Response from starting an OAuth flow.

Parameters:

data (Any)

authorization_url: str

URL to redirect the user to for authorization.

state: str

CSRF state token. Must be passed back in the callback.

class roboto.ai.mcp.UpdateMcpServerAllowedToolsRequest(/, **data)

Bases: pydantic.BaseModel

Admin request to set the tool allowlist for an MCP server.

Keys are tool names, values indicate whether the tool is enabled (True) or blocked (False). Set to None to clear the allowlist (reverts to auto-populate on first discovery).

Parameters:

data (Any)

allowed_tools: dict[str, bool] | None = None

class roboto.ai.mcp.UpdateMcpServerOrgsRequest(/, **data)

Bases: pydantic.BaseModel

Admin request to update the org scoping for an MCP server.

The provided org_ids fully replace the current set. An empty list makes the server global (visible to all orgs). Tokens for removed orgs are cascade-deleted.

Parameters:

data (Any)

org_ids: list[str] = None

New set of org IDs. Empty = global (visible to all orgs).

class roboto.ai.mcp.UpdateMcpServerRequest(/, **data)

Bases: pydantic.BaseModel

Request to update an MCP server registration.

Parameters:

data (Any)

display_name: str | None = None

Updated display name. Alphanumerics, hyphens, and underscores only.

scopes: list[str] | None = None

Updated scopes.