SignalVault
BUILT BY THIS SITE'S EDITORSignalVault is a hosted proxy that logs requests to OpenAI and Anthropic models and blocks or redacts sensitive data in them.
What SignalVault does
SignalVault sits between an application and its model provider. It records each prompt, response and tool call together with the rule decisions that applied, and can block, flag or redact a request that contains sensitive data before the provider receives it. Detection covers email addresses, US social security numbers, card numbers, phone numbers, API keys and tokens.
It is used as a proxy, by changing the client's base URL and adding one header, through a Python or Node.js SDK that wraps the OpenAI and Anthropic clients, or by posting events to a REST API. The proxy and the SDKs stop a blocked request themselves; on the REST API the calling code receives the decision and has to act on it. A mirror mode records requests without sitting in the request path. Stored prompts and responses are encrypted with AES-256-GCM, and the record can be exported as CSV or JSON.
SignalVault: owner, license and hosting
| Fact | Value | Source and date checked |
|---|---|---|
| Owner | Elvar, Stockholm, Sweden. | SOURCE ↗ |
| License | The software is owned by its operator. The terms grant a limited, non-exclusive license to use the service. | SOURCE ↗ |
| Self-hosting | No. The pricing page lists self-hosting as planned. | SOURCE ↗ |
| Hosted regions | Stockholm, Sweden. | SOURCE ↗ |
Each value was read from the vendor's own page on the date shown. How listings are checked
SignalVault plans and prices
SignalVault pricing, checked on . Lowest paid plan: Starter, $49 a month per app. Free tier: No free plan. Each new app gets 100 free requests, no credit card required. Pricing model: A monthly subscription for each app.
| Plan | Price | Included usage | Beyond that | Data retention | Users |
|---|---|---|---|---|---|
| Starter | $49 a month per app | Unlimited logging | – | 30-day retention | – |
| Growth | $149 a month per app | – | – | 90-day retention | – |
| Enterprise | Contact us | – | – | Retention up to 365 days | – |
Each app has its own subscription. Starter covers the core rules (PII, secrets and a token limit). Custom guardrail rules, email and webhook alerts and priority support start at Growth; Enterprise adds custom limits, a contract, a DPA and invoicing. An account has one login on every plan. The pricing page says the product is in early access and that plans can be customised.
As stated on the vendor's pricing page on . A dash means the page does not say.
SignalVault pricing history
The pricing page has been checked daily since . No price change has been recorded since then.
Where SignalVault fits
Teams that call OpenAI or Anthropic directly and need a record of what was sent, with sensitive values stopped before they leave. Integration through the proxy is a base URL change and one added header, and the service is run from Stockholm with its data region there.
Limits of SignalVault
It supports the OpenAI and Anthropic APIs only: the site says Azure OpenAI, other OpenAI-compatible gateways and self-hosted models are not supported yet. There is no free plan beyond 100 trial requests for each new app, and pricing is per app. An account has one login; the pricing page lists team accounts, SSO, team roles and self-hosting as planned. The security page says the product is not SOC 2 certified.
Alternatives to SignalVault
-
Amazon Bedrock Guardrails
Amazon Bedrock Guardrails is a managed AWS service that filters prompts and model responses against content, topic and sensitive-data policies.
-
Google Model Armor
Model Armor is a Google Cloud service that screens LLM prompts and responses for prompt injection, harmful content, sensitive data and malicious URLs.
-
NVIDIA NeMo Guardrails
NVIDIA NeMo Guardrails is an open-source Python library for adding programmable input, dialog, retrieval and output checks to LLM applications.
-
OpenAI Guardrails
OpenAI Guardrails is an open-source Python library that wraps the OpenAI client and runs configurable checks on model inputs and outputs.
-
Presidio
Presidio is an open-source Python framework that detects and anonymizes personal data in text, images and structured data.