Myworkingmemory Specialist Reranking API

Privacy policy

Effective date:

Service status: the direct API and RapidAPI connection have passed synthetic integration tests. Marketplace availability, subscription prices and quotas are shown in the RapidAPI listing. OpenRouter acceptance remains unverified.

This policy covers the Myworkingmemory Specialist Reranking API, operated by Myworkingmemory LLC. The API compares your query with supplied documents and returns relevance scores.

Request content

Queries, documents and scores are used to process the current request. Our application does not save that content to a request database, content log or reusable response cache.

We do not use your request content to train or fine-tune models, build customer profiles, target advertising or create a collection of customer documents. This service has no application analytics or advertising trackers.

Access and identifiers

The API checks an authorization key to control access. Keep that key private. Our gateway does not record the key, your IP address, browser information, cookies or individual request paths in access logs. It does not create customer accounts or retain customer identifiers.

Our gateway returns a randomly generated request identifier to the caller but does not store it. It returns the model's token usage with the response. The current implementation does not persist billing counters or individual request histories.

Processing and retention

Request content is held in working memory while the request is handled. It is not deliberately kept for later requests. Gateway and model-server content logging and reusable prompt caching are disabled in the deployed service configuration. Gateway core dumps are disabled.

This is not a promise that every byte is instantly or securely erased. Runtime buffers and operating-system memory management, including swap storage where enabled, can outlast a request. We do not claim forensic erasure or absolute zero data processing.

Network and platform services

The API route uses Tailscale to carry HTTPS requests to the Mac that runs the model. This route carries requests to the serving Mac. Network infrastructure must process connection information to deliver requests; this policy does not promise that infrastructure providers collect nothing.

These static support, privacy and terms pages are hosted through Vercel. We add no page analytics, tracking scripts, external fonts or account forms to them. Vercel may process ordinary network metadata when delivering a page.

The OpenRouter provider application has been submitted. Acceptance and marketplace availability remain unverified. If the API becomes available through OpenRouter, OpenRouter handles its own platform account, routing and billing processes and receives the content needed to route your request. OpenRouter's handling is separate from our gateway's no-content-storage configuration.

RapidAPI marketplace requests

The RapidAPI connection has passed a synthetic integration test. Marketplace availability is shown in the RapidAPI listing. Requests sent through RapidAPI pass through its marketplace gateway. That gateway can record the full request and response, including headers, query content, documents and returned scores. RapidAPI also handles its own platform accounts, subscriber identifiers, billing and analytics. Our application's no-content-logging configuration does not extend to RapidAPI.

RapidAPI's Gateway documentation describes detailed logs for up to seven days, while its Provider Analytics documentation describes up to 30 days of historical logs containing request and response content. Those descriptions differ. The exact retention period and available logging controls for this listing have not been verified. We make no zero-retention claim for RapidAPI.

Support messages

If you email us, we receive the email address and information you choose to send. Support messages may be retained to handle and follow up on the request. Do not send API keys, passwords, confidential documents or private query content.

For questions about this service or information you voluntarily sent to support, contact dean@myworkingmemory.ai.