SuperContracts MCP plugin

Connect your AI client to the hosted apilabs-ai SuperContracts MCP service. The service evaluates policies and approvals before it invokes protected API or MCP actions. This package contains only client configuration and workflow guidance; it does not contain server code, tokens, or provider credentials.

Setup

1. In apilabs-ai, create a personal MCP token in Auth Vault → MCP Access Tokens.
2. Configure APILABS_MCP_TOKEN in your plugin/client settings. Never commit it.
3. Install the plugin and confirm that supercontracts exposes tools.
4. Start with a read-only request: List my saved contracts with their names and connection IDs.

The configured endpoint is https://mcp.apilabs.ai/mcp using Streamable HTTP.

Included workflows

• setup — secure connection and first-read verification.
• author-contract — create a contract without placing credentials in YAML.
• run-protected-action — validate, execute, and resume an approval safely.
• investigate-run — inspect a saved run and its evidence.

Security

• The MCP token authenticates access to apilabs-ai; it is not a provider credential.
• Tokens can be limited to read, write, execute, and approval permissions. Keep all permissions for normal Cursor use; create a reduced token only for a known limited workflow.
• Store GitHub, Stripe, and other provider credentials in apilabs-ai Auth Vault.
• Never bypass an approval_required result or replace the original request when resuming it.
• Use read-only or validate workflows first when evaluating a new contract.

For direct client setup and local-bridge guidance, see the repository's AI client documentation.