# Inviolet

> Inviolet docs — Intent Detection and Response (IDR) for AI agents. For pricing, sales, demos, or compliance questions, the docs assistant will redirect you to the Inviolet sales agent at https://inviolet.ai/?widget=open&context=docs-redirect.

## Docs

- [Build agents that govern themselves](https://docs.inviolet.ai/index.md): Intent Detection and Response (IDR) for the age of AI agents — every prompt declares an intention, every action runs through policy, every credential is brokered short-lived.
- [Quickstart](https://docs.inviolet.ai/quickstart.md): Five steps, five minutes — from zero to an agent calling a real tool through Inviolet, with the verdict in the live decision feed.
- [Intent Detection and Response (IDR)](https://docs.inviolet.ai/concepts/intentional-access.md): The category Inviolet defines: access governed by why, not just who.
- [The 5 Principles](https://docs.inviolet.ai/concepts/principles.md): The five durable promises Inviolet keeps — the doctrine behind every decision the platform makes.
- [Intent cards](https://docs.inviolet.ai/concepts/intent-cards.md): The unit of authorization declaring what an agent may do, on which columns, under what conditions.
- [Mandates](https://docs.inviolet.ai/concepts/mandates.md): Short-lived JWTs binding a specific actor to an intent card with a specific scope.
- [Cascade classifier](https://docs.inviolet.ai/concepts/cascade.md): Five layers, dispatched per tier, turning a raw prompt into a structured intent with reasoning.
- [Credential broker](https://docs.inviolet.ai/concepts/credential-broker.md): Integrates with your existing PAM so credentials release only when an Inviolet mandate is presented.
- [Decision feed](https://docs.inviolet.ai/concepts/decision-feed.md): Live observability of every intent, mandate, and tool call, with an inline correction loop.
- [The six enforcement layers](https://docs.inviolet.ai/concepts/six-enforcement-layers.md): Where Inviolet sits in your stack — and why each layer matters.
- [The maturity model](https://docs.inviolet.ai/concepts/maturity-model.md): Seven tenets, five levels — how teams climb from observation to structural enforcement, one principle per level, scored against NIST SP 800-207.
- [Tier comparison](https://docs.inviolet.ai/tiers/overview.md): Violet, Deep Violet, Ultraviolet — what each unlocks.
- [Compliance evidence catalog](https://docs.inviolet.ai/compliance/evidence-catalog.md): Exactly what Inviolet records for every AI decision — automated or human — and how each field maps to the EU AI Act, GDPR Article 22, SEC recordkeeping, and California's CCPA ADMT rules.
- [Install the gateway](https://docs.inviolet.ai/guides/install-the-gateway.md): Three install paths — npm SDK, Docker, or Vercel Edge.
- [Connect your first data source](https://docs.inviolet.ai/guides/connect-your-first-data-source.md): Salesforce, Snowflake, or Postgres — the three most common starts.
- [Integrate your identity provider](https://docs.inviolet.ai/guides/integrate-your-idp.md): Connect Okta, Azure AD, or Auth0 to enrich every intent with a verified user.
- [Connect HashiCorp Vault](https://docs.inviolet.ai/guides/connect-vault.md): Bind every grant to a short-lived database credential issued by Vault.
- [Deploy the extension via MDM](https://docs.inviolet.ai/guides/mdm/overview.md): Force-install the Inviolet browser extension across your fleet and push a signed managed-config so each install self-enrolls to your tenant.
- [Microsoft Intune](https://docs.inviolet.ai/guides/mdm/intune.md): Force-install the Inviolet extension and push its managed-config via an Intune Settings-catalog Configuration Profile.
- [Jamf Pro](https://docs.inviolet.ai/guides/mdm/jamf.md): Deploy the Inviolet extension to Macs with a Jamf Pro Configuration Profile carrying two Application & Custom Settings payloads.
- [Kandji](https://docs.inviolet.ai/guides/mdm/kandji.md): Deploy the Inviolet extension to Macs with a single Kandji Custom Profile Library Item.
- [VMware Workspace ONE](https://docs.inviolet.ai/guides/mdm/workspace-one.md): Deploy the Inviolet extension via a Workspace ONE UEM Custom Settings profile — Windows OMA-URI and macOS plist variants.
- [Define your first intention](https://docs.inviolet.ai/guides/define-your-first-intention.md): Move from anonymous observation to identity-aware shadow-mode enforcement.
- [Define your intention policies](https://docs.inviolet.ai/guides/define-your-intention-policies.md): Compose, test, and approve the policies that gate every grant.
- [Generalize from observation](https://docs.inviolet.ai/guides/generalize-from-observation.md): Convert observed query patterns into declarative intentions.
- [Dashboard](https://docs.inviolet.ai/operate/dashboard.md): The operational scoreboard. Reading every panel.
- [Approvals](https://docs.inviolet.ai/operate/approvals.md): The gray zone. Per-intention queues, dual-approver routing, SLA timers, Slack fan-out, step-up auth.
- [Trial-everything](https://docs.inviolet.ai/operate/trials.md): Every Deep Violet and Ultraviolet feature unlocks a 30-day free trial on first use.
- [Migrations](https://docs.inviolet.ai/operate/migrations.md): Running database migrations safely against your Inviolet deployment.
- [Claude for Work](https://docs.inviolet.ai/integrations/llm-hosts/claude.md): Put Inviolet in front of every Claude tool call — Anthropic SDK, Claude Desktop, and MCP.
- [ChatGPT Enterprise](https://docs.inviolet.ai/integrations/llm-hosts/chatgpt.md): Put Inviolet in front of every OpenAI tool call — Assistants, Responses API, and ChatGPT Enterprise actions.
- [Microsoft Copilot Studio](https://docs.inviolet.ai/integrations/llm-hosts/copilot.md): Govern Copilot Studio agents with Inviolet intent extraction and policy.
- [Salesforce](https://docs.inviolet.ai/integrations/data-sources/salesforce.md): Connect Salesforce so every agent query carries a verified intention and matches a column allowlist.
- [Snowflake](https://docs.inviolet.ai/integrations/data-sources/snowflake.md): Connect Snowflake so analytics agents declare an intention for every warehouse query.
- [Postgres](https://docs.inviolet.ai/integrations/data-sources/postgres.md): Connect a Postgres database so transactional agents carry an grant to every query.
- [Network enforcement overview](https://docs.inviolet.ai/integrations/network/overview.md): How Inviolet integrates with SWG / CASB vendors to stop and measure LLM bypass — the two integration directions and which vendor does what.
- [Cloudflare (Gateway + Logpush)](https://docs.inviolet.ai/integrations/network/cloudflare.md): Block LLM bypass with Cloudflare Gateway and stream Gateway DNS logs into Inviolet — the two directions of one Cloudflare integration.
- [Zscaler](https://docs.inviolet.ai/integrations/network/zscaler.md): Connect Zscaler to Inviolet — observe LLM bypass volume (CASB) today; network enforcement (SWG) on request.
- [Netskope](https://docs.inviolet.ai/integrations/network/netskope.md): Connect Netskope to Inviolet — observe LLM bypass volume (CASB) today; network enforcement (SWG) on request.
- [Palo Alto Prisma Access](https://docs.inviolet.ai/integrations/network/prisma.md): Connect Palo Alto Prisma Access to Inviolet — observe LLM bypass volume (CASB) today; network enforcement (SWG) on request.
- [Cisco Umbrella](https://docs.inviolet.ai/integrations/network/umbrella.md): Connect Cisco Umbrella to Inviolet — observe LLM bypass volume (CASB) today; network enforcement (SWG) on request.
- [Microsoft Defender for Cloud Apps](https://docs.inviolet.ai/integrations/network/mcas.md): Connect Microsoft Defender for Cloud Apps to Inviolet — observe LLM bypass volume (CASB) today; network enforcement (SWG) on request.
- [HTTP API](https://docs.inviolet.ai/api-reference/http-api.md): The stable HTTP surface behind every Inviolet SDK — base URL, auth, endpoints, and error codes.
- [POST /v1/intent/evaluate](https://docs.inviolet.ai/api-reference/evaluate.md): The legacy pre-mandate policy path — a quick policy verdict without minting a mandate first.
- [POST /v1/mcp-proxy/call](https://docs.inviolet.ai/api-reference/decide.md): Run the decision engine against a mandate + a normalized action and get the verdict your wrapper should act on.
- [POST /v1/mandate/dispense](https://docs.inviolet.ai/api-reference/dispense.md): Mint a mandate JWT bound to a specific actor + intent card + scope.
- [GET /v1/mandate/jwks](https://docs.inviolet.ai/api-reference/jwks.md): Public-key set used to verify Inviolet-signed mandate JWTs.
- [Intent Events](https://docs.inviolet.ai/api-reference/intent-events.md): Evaluate an LLM tool call, page through history, and verify the resulting grant.
- [Intentions](https://docs.inviolet.ai/api-reference/intentions.md): List, create, and publish the declarative intent patterns Inviolet matches against every tool call.
- [Webhooks](https://docs.inviolet.ai/api-reference/webhooks.md): Receive approval-granted, approval-denied, and audit-export events from Inviolet.
- [Evaluate an agent tool call](https://docs.inviolet.ai/api-reference/intent/evaluate-an-agent-tool-call.md): Extracts the intent, matches against published intentions, evaluates policies, and returns the access decision plus (when allowed) a short-lived grant.
- [List recent intent events](https://docs.inviolet.ai/api-reference/intent/list-recent-intent-events.md)
- [List intentions](https://docs.inviolet.ai/api-reference/intentions/list-intentions.md)
- [Create an intention (status defaults to draft)](https://docs.inviolet.ai/api-reference/intentions/create-an-intention-status-defaults-to-draft.md)
- [Publish an intention (status → active)](https://docs.inviolet.ai/api-reference/intentions/publish-an-intention-status-→-active.md)
- [List approval requests](https://docs.inviolet.ai/api-reference/approvals/list-approval-requests.md)
- [Approve a pending request](https://docs.inviolet.ai/api-reference/approvals/approve-a-pending-request.md)
- [Deny a pending request](https://docs.inviolet.ai/api-reference/approvals/deny-a-pending-request.md)
- [List configured webhook endpoints](https://docs.inviolet.ai/api-reference/webhooks/list-configured-webhook-endpoints.md)
- [Register a new webhook endpoint](https://docs.inviolet.ai/api-reference/webhooks/register-a-new-webhook-endpoint.md)
- [Anthropic SDK](https://docs.inviolet.ai/sdks/anthropic.md): Wrap @anthropic-ai/sdk with Inviolet so every tool call is gated before Claude executes.
- [OpenAI SDK](https://docs.inviolet.ai/sdks/openai.md): Wrap openai with Inviolet so every function-call tool execution is gated.
- [LangChain](https://docs.inviolet.ai/sdks/langchain.md): Wrap any LangChain Tool or Agent so every invocation runs through Inviolet first.
- [LlamaIndex](https://docs.inviolet.ai/sdks/llamaindex.md): Wrap LlamaIndex query engines and tools so retrieval and tool-execution run through Inviolet.
- [MCP client](https://docs.inviolet.ai/sdks/mcp-client.md): Govern any Model Context Protocol client so every tool invocation is gated before the MCP server executes.
- [Fetch wrapper](https://docs.inviolet.ai/sdks/fetch.md): A drop-in replacement for the standard fetch that guards every outbound call.
- [Node SDK](https://docs.inviolet.ai/sdks/node.md): @inviolet/agent-sdk-core — the Inviolet client: mint mandates and guard tool calls.
- [Python](https://docs.inviolet.ai/sdks/python.md): Call the Inviolet gateway from Python over REST — mint mandates and guard tool calls with httpx.
- [REST API](https://docs.inviolet.ai/sdks/rest.md): Hit Inviolet's HTTP API directly — mint mandates and guard tool calls from any language.

## OpenAPI Specs

- [openapi](https://docs.inviolet.ai/openapi.yaml)