AI Agents and API Terms

1. Definitions

• "Agent Key" means a credential beginning with ca_agent_ that you create in Settings, or that is created for you when you approve an OAuth connection, to let an AI Agent act for you.
• "Authority" means, for an Agent Key, its scopes, per-job limit, monthly limit, approval threshold and expiry, as shown in Settings.
• "API Key" means a Customer API or Provider API credential.
• "Connector" means a third-party AI assistant or application connected to your Account through OAuth.
• "Agent Action" means any request made with an Agent Key or API Key.

2. Granting Authority

2.1 Scopes. You choose which scopes an Agent Key has. The scopes are: quote (price estimates and transport modes); jobs:read (see your Jobs and Bids); jobs:write (post and cancel Jobs, spending from your Wallet within limits); bookings:read (see Bookings, tracking and Custody Records); bookings:write (accept Bids, committing funds within limits); wallet:read (see your Wallet balance); and trust:read (check provider credentials). When you connect an AI assistant through OAuth, the consent screen lists each permission the assistant requests with its own checkbox, unticked by default, and CouthActs grants only the scopes you tick and never more than the assistant requested. An Agent Key or OAuth connection created without any scopes receives only the minimum default scopes, quote and trust:read.

2.2 Spending limits. Each Agent Key has (a) a per-job limit, the most a single Job or accepted Bid can cost, including Fees; (b) a monthly limit on the total cost of new Jobs the key posts in a calendar month (UTC); and (c) an approval threshold. Accepting a Bid counts against the per-job limit and the approval threshold but not against the monthly limit, because the funds were already committed when the Job was posted.

2.3 Approvals. Any spending action that costs more than the approval threshold is not carried out until you approve it in CouthActs. You are notified of each request. A request expires if you do not decide within seventy-two (72) hours, and a key may have at most ten (10) requests waiting at once. Approving a request carries it out once, still within the key's limits. Setting the threshold to zero means you approve every spending action yourself.

2.4 What agents cannot do. An Agent Key cannot add payment methods, top up or withdraw from your Wallet, change your Account details, or create other keys.

2.5 Expiry, revocation and changes. You may set an Agent Key to expire, change its limits, or revoke it at any time in Settings. An OAuth connection uses access tokens that expire after one (1) hour and refresh tokens that change on every use and expire after thirty (30) days without use; if a used refresh token is presented again, the connection is revoked and you are notified. Disconnecting a connection in Settings, or through the token revocation endpoint, revokes its tokens. Connections made before short-lived tokens were introduced expire thirty (30) days after that change took effect, after which you will be asked to reconnect. Revoking a key declines all of its pending approval requests. A key stops working if your Account is suspended or closed.

3. Agent Actions Bind You

3.1 Attribution. EVERY AGENT ACTION MADE WITH YOUR AGENT KEY OR API KEY, WITHIN ITS AUTHORITY, IS YOUR ACTION. It binds you as if you had taken it yourself in the Services, including posting a Job (which debits Fees and holds the budget), accepting a Bid (which forms a Transport Contract with the Provider and commits your funds) and canceling a Job.

3.2 No reliance on agent behavior. AI Agents can misunderstand instructions, act on false or malicious content (including prompt injection), hallucinate, or behave unexpectedly. CouthActs carries out Agent Actions as requested and does not review whether an action reflects your wishes. Your protection is the Authority you set; set limits and approval thresholds that you are comfortable with.

3.3 Actions outside Authority. CouthActs is designed to refuse Agent Actions outside a key's Authority. If an action outside the Authority is carried out because of an error in the Services, notify support@couthacts.com within seventy-two (72) hours after you are notified of it, and CouthActs will reverse it where possible (for example, by canceling an unbooked Job and returning the held budget and Fees) and otherwise treat it under the Terms.

3.4 Unauthorized use. You are responsible for keeping keys secret. If you believe a key has been exposed, revoke it immediately. You are responsible for Agent Actions made with a key before you revoke it, except to the extent caused by CouthActs' breach of its security obligations.

3.5 Activity log and notifications. CouthActs records each Agent Action, including the tool used, its status, amount, inputs and result, and shows recent activity in Settings. CouthActs notifies you when a key is created, its limits change, it needs approval, it takes a spending or canceling action, or it is revoked. Review this activity regularly.

4. Third-Party AI Platforms

4.1 Their terms apply. Connectors and AI assistants (such as those provided by Anthropic, OpenAI and others) are operated by third parties under their own terms and privacy policies. When you connect one, CouthActs sends it the data its scopes allow, at your direction.

4.2 No responsibility. COUTHACTS DOES NOT CONTROL AND IS NOT RESPONSIBLE FOR ANY THIRD-PARTY AI PLATFORM, CONNECTOR OR AGENT, INCLUDING ITS OUTPUTS, ADVICE, ERRORS, SECURITY, AVAILABILITY OR USE OF YOUR DATA. Quotes, price guidance and provider information returned through the API are subject to the disclaimers in Section 19 of the Terms.

5. Customer API and Provider API

5.1 Plans and limits. API Keys are subject to the monthly call limits of their plan, shown in the Services: for the Customer API, Starter (1,000 calls, after which calls are refused), Growth (5,000 calls, after which calls are charged at US$0.20 each) and Enterprise (as agreed); and for the Provider API, Basic (500 calls), Professional (5,000 calls) and Enterprise (as agreed). Agent Keys are limited to 120 requests per minute. CouthActs may apply other reasonable rate limits to protect the Services.

5.2 Overage. Growth plan overage is calculated monthly and charged to your Wallet at the start of the following month. Unpaid overage carries forward and may be set off under the Escrow and Payments Terms.

5.3 License. CouthActs grants you a limited, revocable, non-exclusive, non-transferable license to use the APIs and MCP server, in accordance with their documentation, to access your own Account data and act for your own Account. You must not use the APIs to build a competing marketplace, to scrape or resell data, to access other Users' data, or to exceed or evade limits.

5.4 Security. CouthActs stores API Keys and Agent Keys only as keyed hashes and shows each key once, when it is created; a lost key cannot be recovered and must be replaced. Keep keys confidential, do not embed them in public code or client-side applications, and use them only over HTTPS.

5.5 Provider automation. Providers who use the Provider API to bid, accept Jobs or report tracking remain fully responsible for each Bid and Booking and for the accuracy of any location data they report.

5.6 No service level. The APIs and MCP server are provided without a service level commitment and may change. CouthActs will try to give reasonable notice of breaking changes. Features described in documentation that are not yet available are not part of the Services until released.

6. Suspension

CouthActs may throttle, suspend or revoke any Agent Key, API Key or Connector, or block an AI platform, if it reasonably believes it is compromised, abusive, in breach of these API Terms or the Acceptable Use and Prohibited Items Policy, or a risk to the Services or other Users.

7. Liability

Sections 19 to 21 of the Terms apply. Without limiting them, COUTHACTS IS NOT LIABLE FOR LOSSES ARISING FROM AGENT ACTIONS WITHIN THE AUTHORITY YOU SET, FROM THE CONDUCT OF ANY AI PLATFORM OR CONNECTOR, OR FROM THE COMPROMISE OF A KEY YOU FAILED TO PROTECT, and you will indemnify the CouthActs Parties under Section 20.7 of the Terms for claims arising from your AI Agents' actions.

Version History