toolving
ProductHow it worksPricingDocsDownloadsDemosPlaygroundLog inGet an API key

Privacy Policy

Effective date: On publication of this version at toolving.com/privacy.

This policy explains what personal data Toolving collects, why, who receives it, how long we keep it and what you can do about it. Toolving is the hosted Conceiver API, the account workspace and playground at toolving.com, the guided demos, the Python SDK and the Conceiver agent harness (the "Service"). Hydra-Dynamix Inc. ("Hydra-Dynamix", "we", "us") operates it and is responsible for the data described here.

1. What we collect

Your account

  • Sign-in identity. Clerk runs sign-in. Clerk holds your email address, name, profile picture and sign-in methods. Our database stores only the identifier Clerk issues for you, linked to your Toolving account. We do not link accounts by email.
  • API keys. We store each key's name, its ID, when it was created and when it was revoked, and a one-way hash of the secret. We never store the secret itself.

What you send to the API

  • Content. Questions, answers, corrections, teaching pairs, JSON values, imported pair files and training sequences you send through the API, the SDK or the playground.
  • Operation records. Each operation's request, its result and evidence, its status and its credit charge, kept as a receipt so a retried request returns the same result instead of charging or teaching twice.
  • Learning history. The ordered teaching and query events for your account. Available user learning history can be exported; associated entries are removed from that history when an instance retires without storage. Separate instance event records and operation receipts remain.
  • Learned state. Your account's private model state while an instance is live. With persistent storage enabled, the learned state is kept on the service as an account-private artifact after the instance retires. A downloaded copy is under your control; downloading does not by itself delete the service copy.

Billing

  • Your credit balance, reservations, grants, plan and storage measurements.
  • Stripe customer, checkout and subscription identifiers. Stripe collects and holds your card details and billing address; we never receive your full card number.

Technical data

  • Request logs. For each HTTP request: method, route pattern, status, duration and a request ID. The logs do not contain request bodies.
  • Metrics. Request counts and timings by route and status.
  • Browser storage. Clerk sets cookies to keep you signed in. The playground remembers your selected model in your browser's local storage. The Activity panel keeps a browser-tab display of requests and responses for your own model; the underlying API operations are also retained on the server as described above. The demo studio uses a session cookie and session storage to remember its session and selected learner.
  • Fonts. Our pages load fonts from Google Fonts, so your browser sends Google its IP address and a standard browser request.

Guided demos

Guided demos record an audit history of each demo's events for your account, including the inputs you enter, the responses and any errors.

Workflow Workshop

If you choose to participate, we process the workflow information you provide to plan and build the automation with you. Any recording or publication, including promotional footage or tutorials, requires separate consent describing what will be recorded, how it will be used and what may be made public. Signing up for the Service is not recording or publication consent.

2. The Conceiver agent harness

The agent harness runs on your machine and communicates with Toolving. What it sends depends on the host integration and mode.

  • Local history. The harness uses local log pointers to retrieve evidence from your agent host's history. It requests restrictive filesystem permissions for its own history and state files; enforcement depends on your operating system and host configuration. Keeping full logs locally does not mean that all content stays on your machine.
  • Step-based integrations. The harness sends tool names and projected arguments, step and turn outcomes, token counts and compatibility identifiers. Its projection filters credential-bearing fields and recognized secret patterns, replaces absolute paths with a file name and hash, and bounds strings and collections. Projected arguments can still contain content and personal data.
  • Pi integration. Decision requests can include up to 8,000 characters of the latest user request, 2,000 characters of a preceding successful tool result, or 4,000 characters of an error, after recognized secret patterns are redacted. Requests also include the workspace folder name, tool names and compatibility information. Teaching and plan checks can send tool-call arguments and learned plans, including argument values; those values are not guaranteed to be redacted.
  • What we store. Requests, teaching data, outcomes and associated records are retained in your account's learning history and operation records. Redaction is pattern-based and cannot guarantee removal of all credentials, personal data or confidential content. Review what your chosen integration sends before using it with sensitive material.
  • Your key. The harness reads your Toolving API key from the TOOLVING_API_KEY environment variable and does not write it to its configuration files.

3. How we use data

We use the data above to:

  • run the Service: authenticate you, answer requests, learn into your private model, restore instances and keep receipts;
  • bill you and prevent double charges;
  • secure the Service, investigate abuse and diagnose failures;
  • contact you about your account, billing, security and changes to the Service;
  • meet legal obligations and enforce our Terms and Conditions.

We do not sell personal data, share it for cross-context behavioural advertising, or use your content to teach models that serve other accounts. We do not currently send marketing emails. If we introduce them, we will obtain consent where required by law and include an unsubscribe option in each marketing email. You may also opt out by contacting admin@hydradynamix.com. Opting out of marketing will not stop essential account, billing, security or policy-change notices.

4. Who receives data

  • Service providers that process data for us under contract:
    • Clerk, for sign-in and account profiles;
    • Stripe, for payments and invoices;
    • Hetzner, which hosts the API servers and database in Hillsboro, Oregon, USA;
    • Vercel, which serves the website and routes requests to the API;
    • Cloudflare, which runs the bot check shown during sign-up;
    • Google, for web fonts.
  • Legal and safety. Authorities or other parties when the law requires it, or when needed to protect people, the Service or our rights.
  • Business transfers. A buyer or successor in a merger, acquisition or sale of the Service, subject to this policy.

5. How long we keep data

  • Account data, retained instance events, receipts, saved training sequences and stored artifacts: until you ask us to delete them. Cancelling a subscription does not delete them, and we do not delete them on a schedule.
  • Live instance state without storage: discarded when the instance retires, along with its associated entries in the user learning-history export. Separate instance event records and operation receipts remain; retiring an instance is not an account-data deletion request.
  • Request logs: up to 14 days. Metrics: up to 15 days.
  • Backups: service recovery backups include the service database, saved model artifacts and retirement receipts. These can include retained account and operation data; they are not a separate backup of your local agent logs. A local recovery backup is scheduled daily at 05:30 UTC. We retain one completed local backup and delete the previous one after its replacement succeeds. Hetzner retains up to seven rotating server backups and removes the oldest when a new backup exceeds that limit. After data is deleted from the live service, subsequent successful backups reflect the deletion; older copies remain until replaced by this rotation. Retention depends on successful backup runs, rather than a guaranteed number of calendar days. Separately retained deployment or incident-recovery backups are deleted within 30 days of creation. Deletion from the live service does not immediately remove data from existing backups. Before restoring a backup to active service, completed deletion requests must be reapplied so that restoration does not reintroduce deleted personal data.
  • Billing records: records needed for Canadian tax and accounting obligations are generally retained for six years from the end of the last tax year they relate to, including after account deletion. Records may be retained longer where required by law, an audit, an objection or an appeal. This period applies to required financial records, not automatically to your model content or session data.
  • Clerk and Stripe keep their own records under their own policies.

6. Your choices and rights

  • Export. Download your learning history from Settings and your training sequences from Training sequences in your workspace.
  • Keys. Revoke API keys from API keys at any time.
  • Profile. Change your name, email and sign-in methods under Settings.
  • Deletion. Ask us to delete your account and its data at admin@hydradynamix.com. We will confirm the request comes from the account holder before acting.
  • Other rights. Depending on where you live, you may have rights to access, correct, delete or port your data, to object to or restrict processing, and to complain to a data protection authority. Contact admin@hydradynamix.com to use them. In British Columbia, you can also contact the Office of the Information and Privacy Commissioner for BC. Where EEA or UK data protection law applies, these rights include withdrawing consent at any time for processing based on consent, without affecting the lawfulness of earlier processing, and complaining to your local supervisory authority or the UK Information Commissioner. Where applicable US state privacy law provides them, you may also request information about the categories of personal data collected and disclosed, use an authorized agent, and appeal a refused request. Submit requests or appeals to admin@hydradynamix.com; we may verify your identity and an agent's authority and will respond within the time required by applicable law. We will not unlawfully discriminate against you for exercising privacy rights. Rights are subject to applicable legal conditions and exceptions.
  • Do Not Track. The Service does not track you across other websites, so it does not respond differently to browser Do Not Track signals.

7. Security

API keys are stored as one-way hashes. Each account's learning is isolated to that account. Traffic to the Service is encrypted in transit. The agent harness uses local history and sends the data described in section 2; its redaction does not cover every value or integration. No system is perfectly secure; if you think your account or a key has been compromised, revoke the key and contact admin@hydradynamix.com.

8. Children

We do not set a fixed minimum age. Where a child cannot provide valid consent under applicable law, consent must come from a parent or legal guardian. If you believe we have collected a child's personal data without required consent, contact admin@hydradynamix.com so we can investigate and address the request.

9. International transfers

We host the Service in the United States. If you use it from elsewhere, your data is transferred to and processed in the United States. Our providers publish data-processing terms, including Clerk, Stripe and Vercel. Hetzner makes its data-processing agreement available through its customer account. These published documents do not by themselves establish which transfer safeguards cover your data. Verification of the agreements and transfer safeguards applicable to Toolving is not yet complete; we cannot currently confirm a specific EEA or UK transfer mechanism for the Service. Contact admin@hydradynamix.com for information about the review. Accepting this policy is not a substitute for safeguards required by applicable law.

10. Changes to this policy

We will post updates here with a new effective date. For material changes we will also notify account holders by email to the address associated with your account.

11. Contact

Hydra-Dynamix Inc.

3200-650 West Georgia Street

Vancouver, BC, V6B 4P7, Canada

Email: admin@hydradynamix.com

Learning that carries forward.
DocsQuickstartDownloadsSDK releases ↗TermsPrivacy

Powered by Hydra-Dynamix Inc.