SpiderGate adds a key vault contributors can fill without exposing a secret

TALLINN, Estonia — July 18, 2026 — SpiderIQ, the company behind the SpiderGate LLM gateway, today announced details of the key vault it operates for SpiderGate customers, because teams running AI agents routinely need capacity from provider accounts they do not own. Since July 2026 the vault has let a third party contribute a provider API key through a single-use invite link without the requesting administrator ever seeing the secret [source: SpiderGate product documentation, https://gate.spideriq.ai/docs/dashboard/vault]. Contributed keys are encrypted at rest and pooled with round-robin failover, so a request continues on the next healthy key when one rate-limits or stops working.

The vault addresses a problem common to teams running AI agents: the person who owns a provider account is frequently not the person operating the gateway. Keys are consequently pasted into chat threads and copied into environment files across several machines. In SpiderGate, an administrator sends an invite, the contributor opens it on their own machine and submits the key there with no login and no account, and the administrator afterwards sees only a masked preview of the first and last four characters [source: SpiderGate contributor-invite flow, product documentation, https://gate.spideriq.ai/docs/dashboard/vault].

As of 18 July 2026 the vault held 75 active provider keys across 15 providers, 50 of them pooled [source: SpiderIQ internal key registry, as of 2026-07-18]. Over the preceding 30 days, 68 distinct keys served 169,581 gateway requests through the rotation [source: SpiderGate request log, 30 days to 2026-07-18]. Thirty-one keys have been added through the contributor invite flow [source: SpiderIQ credential-invite records, as of 2026-07-18].

"The interesting constraint was not storing a key safely, it was accepting one from somebody who is right not to trust you with it," said Martin Shein, founder of SpiderIQ. "So the design goal became narrower than encryption. It became: the key must be usable by the gateway and never visible to the person who asked for it."

Keys are classified by how they bill, and the classification governs pooling. Free-tier and flat-fee subscription keys join the shared pool. Metered pay-as-you-go keys never do, a restriction enforced by a database trigger, a check constraint, and rejection at the write API, so that one tenant cannot draw on another tenant's metered billing [source: SpiderGate Vault API reference, https://gate.spideriq.ai/docs/reference/vault].

Key selection proceeds by health first, then least-used, then least-recently-used [source: SpiderGate Vault API reference, https://gate.spideriq.ai/docs/reference/vault]. A key that returns three consecutive errors is marked unhealthy and dropped from selection, and the following request completes on the next healthy key. When a credential fails, the original contributor receives a re-authentication link and replaces it in place, preserving the key's identifier, limits and history.

Per-key spend is derived from the gateway's own request log rather than from provider billing APIs, so cost is attributable to the individual key that served each request.

"Most key managers stop at storage," said Martin Shein, founder of SpiderIQ. "The part that decides whether you can share capacity across a team is the billing classification, and that is why a metered key is refused entry to the pool in three separate places rather than one."

The vault is available to SpiderGate customers now on the Vault tab of the SpiderGate dashboard, and is documented at https://gate.spideriq.ai/docs/dashboard/vault.

About SpiderIQ

SpiderGate is the LLM gateway built for agencies: keys live encrypted in a Vault you contribute to without exposing them, and a shared pool with least-busy failover keeps requests flowing when a key rate-limits or dies. On top sits Studio — a chat workbench where you switch between models mid-conversation, with your files, notes, and saved prompts organized by project.

SpiderGate's routing engine is built on LiteLLM, an open-source router by BerriAI. Provider and platform names referenced above are the trademarks of their respective owners and imply no affiliation or endorsement.

Media contact

Press Office, Media Relations, SpiderIQ press@spideriq.ai

###