What a provider may receive
The data depends on the template and selected model or tool. It may include:- prompts and workflow inputs;
- uploaded text, images, documents, or workbook fields used by a step;
- upstream outputs needed by a later workflow step;
- request settings and the selected model;
- technical metadata needed to execute, troubleshoot, secure, and account for the request.
Provider retention and logging
Third-party providers may log, retain, or otherwise process submitted data according to their own terms, privacy notices, data-processing agreements, regional policies, and account-level controls. Their rules can differ by model, endpoint, account type, and region. Before using sensitive data:- identify which models, APIs, or tools the workflow will call;
- review the applicable provider terms and retention controls;
- remove data that is not required for the workload;
- use organization-approved models and providers;
- avoid storing temporary signed artifact URLs in long-lived logs.
CogFoundry’s training position
CogFoundry does not train CogFoundry models or third-party models on Customer Content unless the user or organization explicitly authorizes it. This does not replace the separate data-handling terms of a provider used for execution.Logs and support
CogFoundry may retain service logs and execution metadata for security, reliability, billing, troubleshooting, and support. Do not include real tokens or complete workbook payloads when sharing diagnostic output. Remove secrets before opening a support request or GitHub issue.LoomLoom security notes
Review token, endpoint, workbook, artifact URL, confirmation, and retry safeguards.