MCP server

Connect Codex, Claude Code or another MCP-compatible AI agent to launch and track your audits.

Cabfy's MCP server lets an AI agent launch and track the four pillars separately: Agent Readiness, Agent Commerce, Agent Performance and Web Findability. It uses Streamable HTTP and a Bearer API key.

Connect

text

https://api.canagentsbuyfromyou.com/mcp/

Streamable HTTP MCP endpoint

Keep the key in the agent configuration

Never put your API key in a prompt, a Git repository or a tool argument. Use an environment variable or your agent's secret manager.

Connect with Codex

Set the key in your terminal, then add this block to ~/.codex/config.toml. Codex and its IDE extension share this configuration:

bash

export CABFY_API_KEY="ar_live_…"\ncodex mcp list

Prepare the key and verify the connection

text

[mcp_servers.cabfy]\nurl = "https://api.canagentsbuyfromyou.com/mcp/"\nbearer_token_env_var = "CABFY_API_KEY"

~/.codex/config.toml

Connect with Claude Code

Set CABFY_API_KEY in your environment, then add this .mcp.json file at the project root. Claude Code resolves the variable when the connection starts:

bash

export CABFY_API_KEY="ar_live_…"\nclaude mcp list

Prepare the key and verify the connection

json

{\n  "mcpServers": {\n    "cabfy": {\n      "type": "http",\n      "url": "https://api.canagentsbuyfromyou.com/mcp/",\n      "headers": { "Authorization": "Bearer \${CABFY_API_KEY}" }\n    }\n  }\n}

.mcp.json

In Claude Code, use /mcp to confirm that the server is connected.

Available tools

Tool	Usage
run_agent_readiness_scan	Launch the general Agent Readiness scan
run_agent_commerce_audit	Launch the Agent Commerce audit
run_agent_performance_audit	Launch Agent Performance after auditing the domain
run_web_findability_audit	Launch Web Findability after auditing the domain
get_agent_audit_result	Read an audit by ID
get_agent_performance_result	Read the latest Agent Performance result
get_web_findability_result	Read the latest Web Findability result
get_account_usage	Read the account plan and quotas

Launch and track an audit

Ask your agent to call run_agent_readiness_scan with the domain URL. The server returns an ID and pending status:

json

{ "business": "https://my-store.com" }

Arguments for run_agent_readiness_scan

The agent should then call get_agent_audit_result with audit_id until the status is done. Launch tools are asynchronous and may consume account quota.

Specialized pillar prerequisites

Agent Performance and Web Findability require an existing Agent Readiness or Agent Commerce audit for the domain. Quotas and the active-site limit are enforced server-side.