Terms

The agreement

These terms are an agreement between you and Since.dev. You accept them by creating an account, requesting an API key, or using the service. If you do not agree, do not use it.

You must be at least 16 and legally able to enter a contract. If you are using Since.dev for an organisation, you confirm that you are authorised to bind it, and “you” means that organisation.

What the service does

Grant is an AI operating assistant for your connected projects. It leads a preset team of AI teammates that review authorized GitHub, AWS, OpenAI, Anthropic and World State evidence, and documents you upload, in check-ins and when asked. Each completed check-in saves a brief, and its findings are saved and prioritized. Grant answers questions, and people and connected agents can record decisions, owners and revisit dates on findings. The team covers CI, cloud and AI provider costs, runtime metrics, existing security findings, dependency compatibility and repository configuration, and on Studio and Enterprise also access reviews and the security of your own agents. For the leaver sweep, on Studio and Enterprise, the team also reads Slack member lists, and OpenAI and Anthropic organization members and API key metadata. Grant can remind owners about findings, track incident follow-ups and send alerts you set. Projects contain the repositories, paths and cloud resources you select. A completed check-in or answer can rest on incomplete evidence coverage.

Since.dev takes a standing interest, matches it to public sources, checks those sources on a schedule, detects changes, and reports them as typed differences: the field that moved, what it was, what it is now, the source, and when it was read. An interest is registered by typing a sentence, by an agent registering a typed watch, by asking Grant on a paid plan, or by subscribing to a supported public source. Changes reach you over HTTP, through the MCP server, or as signed webhooks. AI may help interpret a request and assess whether a change matters. Your watch threshold controls webhook delivery, and you can set a separate email threshold or turn email off entirely.

Grant authority and connected accounts

Workspace owners and administrators choose connected sources, project scopes and delivery destinations, may connect their own AI provider account and model on a paid workspace, and can pause Grant’s work. You must have permission to provide this access and the content processed through it. Check-ins only read the sources they review. The exception is Vera, who sends test prompts to the agent endpoints the workspace owner connects for testing, in check-ins and when asked. Grant and its teammates do not change cloud resources, enable scanners, rerun CI, merge code or deploy projects. On a paid plan, Grant can create or change watches and run fact checks in a conversation, directly or through a teammate, within the permissions of the person or agent asking. Draft repair PRs come only from the compatibility repair flow below. Its permissions are a separate authorization and are not expanded by a project grouping or a conversation request.

Grant is software, not an appointed company officer. You retain authority over business decisions and actions. A recorded decision, such as marking a finding addressed or deferred, does not verify that an underlying issue has been fixed. Revoke access, pause work or archive projects through the product. Pausing Grant, a teammate, a project or a connection stops the queued and running work it covers. Archiving a project stops its check-ins, and revoking a connection stops reads through it. A request already sent to a provider may complete and incur its charges.

Slack installation and identity linking connect messages addressed to Grant with a permitted workspace conversation. Answers to questions asked in a Slack channel, or with /grant in a conversation Grant has not joined (including a direct message between people), are posted to that conversation for everyone in it and limited to the projects the connection is configured for. Answers in a direct message with Grant reach only the linked person who asked. Signed webhook destinations receive signed notices that link to saved results, and on Pro and up email destinations send the same notices to the workspace members they list who keep Grant emails on. A check-in notice also carries its title and any urgent findings’ titles and severities, never evidence, and an alarm, deploy check or risky change notice also lists the urgent findings its check filed, by title and severity. A Monday brief notice also carries Harper’s summary of the brief, and an agent security notice carries each endpoint’s name and score in place of finding titles. Alarm and risky change notices include the evidence behind them in the destinations you choose, including Slack channels, email and webhooks. Treat what they say about a cause as a lead to check, not a verified finding. Delivery retries reuse saved results and do not rerun Grant’s work. A send with an uncertain outcome may require an explicit retry decision.

Alerts you set are checked when the teammate who reads that evidence finishes a run, such as a check-in. They fire once each time a threshold is crossed, or once for each new critical alert or alarm, and do not replace your own monitoring. Owner reminders and incident follow-ups link to the finding, where people record decisions. When Grant sees proof that an incident follow-up is done, he tells you, but only a person marks it done.

Repository access and repair proposals

Connecting a repository authorizes analysis within the selected GitHub access and product settings. Write-capable access can create branches, commits, and draft PRs when you explicitly request that work or enable it for supported repair types. You must have authority to grant access to the repository and its contents.

Repairs are proposals for review. You remain responsible for evaluating the patch, running relevant tests and project evaluations, and deciding whether to merge or deploy. Passing CI does not guarantee production safety. The service does not automatically merge, deploy, modify production secrets, or disable branch protection. Unsupported changes may result in migration guidance instead of a patch.

What Since.dev does not promise

This section is the important one.

• It is not exhaustive. Coverage is limited to supported sources, connected accounts, selected resources and available evidence. Missing permissions, disabled scanners, unavailable recommendations, sampling and incomplete pagination can leave gaps. A security finding review is not a penetration test or compliance certification.
• Silence is not a guarantee. Hearing nothing means no notification was delivered. It does not prove that every source was checked, that a project is healthy or that nothing changed. Inspect work status and evidence coverage before drawing conclusions.
• It is not instant. World State polling and scheduled Grant check-ins have separate cadences. Source delays, event delivery, execution and concurrency limits, pauses and provider limits affect when work completes.
• Delivery is at-least-once. The same change may reach you more than once. Every change carries a stable idempotency key so that you can deduplicate, and you are expected to.
• Unknown is not unchanged. A verdict of unknown means there is no known value to compare against. Treating it as confirmation is the most expensive mistake you can make with this API, and it is your responsibility not to.
• Sources can fail. Registries, feeds, and pages change structure, go down, and restrict automated access. When that happens the watch is marked degraded or uncoverable and you are told. A failure is never reported as “no change”.
• It is not advice. Nothing Since.dev sends is legal, medical, investment, or professional advice. AI explanations can be wrong. Cost estimates and potential savings are not guaranteed outcomes; inspect the evidence and assumptions before acting.

Do not rely on Since.dev as the sole safeguard for a decision with legal deadlines, safety consequences, or irreversible effects. Keep a second check on anything that matters that much.

Acceptable use

You agree not to use Since.dev:

• to monitor an individual’s location, movements, schedule, presence, relationships, social activity, household, or private life;
• to assemble a profile or dossier about a private individual, however framed;
• to harass, stalk, intimidate, locate, or confront any person;
• to make decisions about a person’s employment, credit, housing, insurance, or eligibility for a benefit, for which this service is not a permissible source;
• to bypass a paywall, login, or access control; this does not prohibit connecting accounts and resources you are authorized to access through the supported integrations;
• to circumvent a source’s robots.txt, rate limits, or terms of use;
• to resell, redistribute, or republish source content retrieved through the service, or to build a substantially similar competing service from it;
• to probe, overload, or interfere with the service, or to exceed your plan’s limits by holding multiple accounts.

Since.dev refuses requests that fall inside these categories and may suspend an account for misuse. This is not negotiable and it is applied regardless of the stated purpose.

Your account and your keys

Web, Slack and MCP access use the same Grant capabilities, subject to the identity and permissions of each caller. An owner or administrator must explicitly grant a coding agent its project scope and permitted actions. Questions a permitted agent asks use the workspace’s Since AI credits or, when the workspace uses its own AI provider, that provider connection, and fact checks count toward the monthly verification limit. A connected agent cannot expand its own access, and sharing a repository does not also share cloud evidence or Grant’s saved project results.

Keep your credentials and API keys secret. A Since API key is shown once and only a digest is stored, so we cannot recover one for you, but you can revoke and replace a key at any time. You are responsible for everything done with your keys, including by an agent you operate. Tell us promptly at support@since.dev if you suspect unauthorised access.

Plans, limits, and billing

• Paid plans are billed in advance through Stripe, monthly or annually as you choose, and the current price is shown at checkout before you agree to it. An annual plan is charged once for the year. Fees exclude taxes, which are added where they apply.
• Every plan includes Since AI credits each UTC month: 1,000 on Free, 25,000 on Pro and 80,000 on Studio, and the amount in the contract on Enterprise. Credits measure the AI work Since.dev does for Grant, his team and repository repairs; they reset on the 1st (UTC) and do not roll over. Free and Enterprise workspaces stop using Since AI when their credits run out, until the 1st or a plan change. On Pro and Studio, work past the included credits continues and is billed automatically as extra credits at $2.00 per 1,000 on Pro and $1.50 per 1,000 on Studio, with no cap and nothing to turn on. Extra credits are billed after each UTC month closes on a separate Extra credits invoice, charged to the card on your subscription. Each month is priced at the higher of Pro and Studio that the workspace was on during it, with Studio above Pro, and that plan’s included credits and rate apply to the whole month. Moving from Studio to Pro or to Free during a month, for example, prices that month at Studio, and so does moving from Pro to Studio. A month that ends under an Enterprise contract is billed as that contract says. If you cancel, or move to Free, during a month, the extra credits you used are still billed after the month closes, to the card from your last subscription. If you delete a workspace that was on Pro or Studio during the month, or the account that owns it, its extra credits so far that month are invoiced and charged to the card on file before the deletion, together with any Extra credits invoice still unpaid. If that charge fails, the deletion still goes ahead, and we keep only the billing record Stripe needs to collect it (the billing email, the card the invoice is charged to and the unpaid invoice), until that invoice is paid or we cancel (void) it, then delete it. A month with less than $1.00 of extra credits is not billed. While a subscription payment is past due, or an Extra credits invoice is still unpaid 7 days after it was issued, Since AI on Pro and Studio stops at the included credits until the payment goes through; after an Extra credits invoice is paid, Since AI resumes within a day. Work on your own AI provider uses no credits and keeps running. We email the workspace owner when the included credits run out and each time extra credits pass a multiple of the plan price. Rates and allowances may change with notice under these terms.
• Plans limit active or paused projects, workspace members, how often scheduled Grant check-ins run, World State watched subjects, and monthly verification calls. A project may contain several repositories, part of a monorepo or only cloud resources. Archiving it releases project capacity and stops its operational work without deleting its history. A watch counts once per distinct subject across its aspects and attached agents. A verification counts per call, rather than per fact. Free repository analysis uses the workspace’s Since AI credits. Grant Free includes one project and, per UTC month, one bounded baseline and five follow-up check-ins; on Free, Grant answers questions from saved results. These baselines, follow-ups and answers use the monthly Since AI credits. The eligible draft repair PR runs even when the credits are used up; its AI work still counts toward them. Technical retries do not consume another baseline or follow-up. Paid Grant, repository analysis and repairs run on Since AI by default; an owner or admin may instead use a supported AI provider of their own, and that provider bills model usage directly under your agreement with it. Source providers may also charge for API reads. Paid plans include Grant and his team (Vera, who tests your own agents, is on Studio and Enterprise only) with Since AI credits each month in place of any per-question budget; each run is bounded by internal token and source-read limits, and work by concurrency and request limits. Paid repository work has no repair-count quota or dependency count limit. There are no credit packs, top-ups or rollover, and no automatic overage on verification calls. Extra credits are the only automatic charge beyond your plan price. Webhook deliveries and their retries are never metered or billed.
• Pro includes five projects, three members and scheduled Grant check-ins up to daily. Studio includes fifteen projects, ten members and check-ins up to every six hours. Both include event check-ins and on-demand conversation in which Grant can ask its team for new work. Enterprise limits and support follow the signed contract.
• World State watches retain their separate cadence and allowances. A watch runs at whichever is slower, your plan’s floor or the aspect’s own default, so on most aspects the default is what you get and the plan makes no difference to how often that watch is checked. Grant check-in schedules do not slow World State polling. Treat GET /v1/usage as the source of truth for your account.
• Email to a person is rate limited per account, so one large change across many watches cannot become many emails. A change held back this way is still detected, still delivered to your agents, and still recorded on the watch with the reason it was held.
• Rate limits protect the service and the sources it reads. Sustained excess may be throttled. Talk to us before you need a higher limit rather than after.
• Your login includes one Free owned workspace. Each additional owned workspace requires its own paid subscription. Joining an invited team does not use your Free workspace. Checkout applies to the selected workspace.
• You can cancel at any time. Access continues to the end of the period you have paid for, and we do not refund partial periods except where the law requires it. Downgrading changes future limits and never deletes your history. Extra credits you used before you cancel are still billed after the month closes, at the higher of Pro and Studio you held that month, and deleting a workspace that was on Pro or Studio that month, or the account that owns it, charges its extra credits so far first, as described above. If usage exceeds the applicable project, watch, Free dependency, or member limits, monitoring, AI analysis, and repair work pause until you reduce usage or choose a suitable plan. Saved data and cleanup controls remain available. An additional workspace without a paid subscription retains saved-data access and cleanup controls while monitoring, AI analysis, and repair work stay paused.
• Moving existing customers to Grant preserves existing subscriptions, billing periods, histories and any legacy benefit that would otherwise be reduced. Project grouping does not expand repository repair consent or an external agent's access. Paying or upgrading does not clear a manual pause.
• Quotas and prices may change with notice. Material reductions or increases will be announced by email before they take effect, and you may cancel rather than accept them.
• The free tier is offered as-is, may change, and may be withdrawn. We will give notice before withdrawing it.

Your content

You keep ownership of your content, including project context, repair instructions, conversations, decisions and watch statements. You grant Since.dev the licence needed to operate the service for you, which is to store it, perform the authorized analysis, maintain history and deliver results to your chosen destinations. Provider processing follows the connected account's applicable terms and settings. Deletion and necessary retained records are described in our privacy policy. We do not train models on your content.

Source content

Changes and notices quote short factual excerpts from public sources and link to the original. Rights in that material remain with the publisher. Since.dev is a change detector, not a content republisher, and your use of quoted material is your responsibility.

Availability and changes to the service

We aim for high availability but do not guarantee uninterrupted operation. Maintenance, source outages, and upstream rate limits can delay checks. We may add, change, or remove features. Where a change removes something you depend on, we will give reasonable notice through the API changelog or by email.

Warranties

The service is provided “as is” and “as available”. To the maximum extent permitted by law, Since.dev disclaims all implied warranties, including merchantability, fitness for a particular purpose, and non-infringement. We do not warrant that the service will be uninterrupted, error free, or that it will detect any particular change. Some jurisdictions do not allow these exclusions, in which case they do not apply to you.

Liability

To the maximum extent permitted by law, Since.dev is not liable for indirect, incidental, special, or consequential loss, for lost profits or revenue, or for any loss arising from a change Since.dev did not detect, did not classify as material, or did not deliver. Our total liability in any twelve-month period is limited to the amount you paid us in that period. Nothing here excludes liability for fraud, death or personal injury caused by negligence, or anything else that cannot lawfully be excluded.

Indemnity

You will indemnify Since.dev against claims, losses, and reasonable costs arising from your use of the service in breach of these terms, in particular a breach of the acceptable-use section.

Termination

You may close your account at any time from Settings. We may suspend or terminate an account that breaches these terms, with notice where that is practical and immediately where a person may be at risk or the service is being harmed. On termination your right to use the service ends; the sections on content, liability, indemnity, and governing law survive.

Governing law and disputes

These terms are governed by the laws of the State of Delaware, United States, without regard to its conflict-of-laws rules, and the courts of that jurisdiction have exclusive jurisdiction over any dispute.

If you are a consumer, these terms do not deprive you of the protection of the mandatory law of the country where you live, or of the right to bring proceedings there.

Before filing anything, write to legal@since.dev. Most disputes are a misunderstanding about coverage, and we would rather fix it than argue about it.

General

• Entire agreement. These terms and the privacy policy are the whole agreement between us about the service, except where a separately signed order form, services agreement or data-processing agreement applies. Those documents govern their agreed subject matter and any expressly stated precedence.
• Severability. If a provision is unenforceable, the rest stands.
• No waiver. Not enforcing a term once does not waive it.
• Assignment. You may not assign these terms without our consent. We may assign them to a successor in a merger or sale of the business, on notice to you.
• Force majeure. Neither party is liable for a failure caused by something outside its reasonable control.
• Notices. We contact you at the email address on your account. You contact us at the addresses on this page.

Changes to these terms

Material changes will be announced by email before they take effect, and the date at the top of this page will change. Continuing to use the service after that is acceptance. If you do not accept a change, cancel before it takes effect. Legal questions: legal@since.dev.