Solutions

Everything Argent connects to

AI model providers, coding agents, identity systems, secrets vaults, policy engines and the tools around them. One page each on what Argent does with it and how it connects.

Showing all 138 connects to
Direct APIOpenAIEvery OpenAI call checked, redacted, budgeted and signed. The API key stays in your vault.View Direct APIAnthropicClaude calls from apps and coding agents, governed and recorded. The Anthropic key stays in your vault.View Direct APIGoogle GeminiGemini Developer API calls brokered: model read from the URL, rules, redaction, budgets, signed record.View Direct APIMistralMistral API calls governed, with the Mistral key held in your vault.View Direct APIDeepSeekRefuse direct DeepSeek calls, swap them to an approved model, or run DeepSeek in Azure Foundry under your rules.View CloudAzure OpenAI and FoundryAzure OpenAI and Foundry deployments resolved to the real model, swapped, redacted, budgeted and signed.View CloudAmazon BedrockBedrock Converse and InvokeModel governed, streams included. Argent signs with SigV4; callers hold no AWS key.View CloudAmazon Bedrock MantleBedrock's OpenAI- and Anthropic-compatible endpoints, with the Bedrock API key held by Argent.View AggregatorOpenRouterOne OpenRouter key, many models, and your approved list, redaction and budgets on every call.View AI gatewayPortkeyKeep your Portkey gateway. Argent adds agent identity, approved models, redaction, budgets and a signed record.View AI gatewayKong AI GatewayKeep Kong AI Gateway. Argent adds agent identity, approved models, redaction, budgets and a signed record.View AI gatewayCloudflare AI GatewayKeep Cloudflare AI Gateway. Argent adds agent identity, approved models, redaction, budgets and a signed record.View Coding toolGitHub CopilotCopilot model traffic inspected through the proxy while each user keeps their own Copilot sign-in.View Coding toolKiroKiro model calls through the proxy: approved operations only, user sign-in kept, metered and signed.View Self-hostedOllamaSelf-hosted Ollama under the same rules, redaction, budgets and signed record. No key to manage.View Self-hostedvLLMSelf-hosted vLLM under the same rules, redaction, budgets and signed record. No key to manage.View Self-hostedNVIDIA NIMSelf-hosted NVIDIA NIM under the same rules, redaction, budgets and signed record. No key to manage.View Self-hostedLM StudioSelf-hosted LM Studio under the same rules, redaction, budgets and signed record. No key to manage.View Self-hostedHugging Face TGISelf-hosted Hugging Face TGI under the same rules, redaction, budgets and signed record. No key to manage.View Any providerAny OpenAI-compatible endpointAny service that speaks the OpenAI API, governed with its key in your vault.View AI clients and agentsClaude CodeEvery Claude Code call and tool action runs through Argent: a named, owned agent, redaction, spend caps and a signed record.View AI clients and agentsClaude Code in VS Code and JetBrainsThe Claude Code extensions share the CLI's settings, so one Argent setup governs the terminal and the IDE.View AI clients and agentsOpenAI Codex CLICodex model traffic and tool actions pass through Argent, attributed to an agent your developer owns.View AI clients and agentsOpenAI Codex IDE extensionThe Codex IDE extension reads the same config as the CLI, so Argent governs both from one setup.View AI clients and agentsGemini CLIGemini CLI calls are checked, redacted, priced and signed by Argent, with hooks on every tool action.View AI clients and agentsGitHub Copilot in VS CodeCopilot Chat in VS Code runs through the Argent forward proxy while users keep their own Copilot sign-in.View AI clients and agentsGitHub Copilot in JetBrainsCopilot in JetBrains IDEs uses the IDE proxy and the Argent certificate, so its calls land under policy.View AI clients and agentsGitHub Copilot in Visual StudioCopilot in Visual Studio follows the Windows proxy, so Argent can see, check and record its calls.View AI clients and agentsGitHub Copilot CLICopilot CLI model calls go through the Argent proxy and every tool action asks Argent first.View AI clients and agentsCursorArgent hooks check every Cursor prompt, shell command, file read and MCP call before it runs.View AI clients and agentsCursor Agent CLIcursor-agent shares Cursor's hooks, so Argent checks every prompt and tool action in the terminal too.View AI clients and agentsKiro CLIKiro CLI traffic is inspected, metered and signed by Argent while the user's own Kiro sign-in stays in place.View AI clients and agentsKiro IDEThe Kiro IDE gets the same Argent hooks, proxy and certificate as the CLI, under the same policy.View AI clients and agentsClinePoint Cline's provider at Argent and add Argent hooks: model calls and tool actions both come under policy.View AI clients and agentsRoo CodeSet Roo Code's provider to your Argent route and every call is attributed, checked and recorded.View AI clients and agentsKilo CodeKilo Code's provider setting points at Argent, so its calls in VS Code, JetBrains and the CLI land under policy.View AI clients and agentsContinueA Continue config.yaml that names your Argent route puts every chat and edit under Argent policy.View AI clients and agentsJetBrains AI AssistantBring your own model key to Argent and JetBrains AI Assistant calls are checked and recorded.View AI clients and agentsJunieA custom model profile sends Junie's calls to Argent, attributed to an agent with a human owner.View AI clients and agentsOpenCodeOpenCode runs through Argent by proxy, hooks or base URL, with one argent connect command.View AI clients and agentsDevin for TerminalArgent hooks check every Devin CLI prompt and tool action before it runs, and sign the decision.View AI clients and agentsZedZed's agent panel calls your Argent route, so every request is attributed, checked and signed.View AI clients and agentsAiderAider takes an OpenAI-compatible base URL, so pointing it at Argent puts every call under policy.View AI clients and agentsGooseGoose's provider points at Argent, so its model calls are attributed, redacted, capped and recorded.View AI clients and agentsTabnineArgent sits between your Tabnine Enterprise server and its models, as a self-managed model endpoint.View AI clients and agentsSourcegraph CodyArgent sits on Cody Enterprise's model leg, so the calls your Sourcegraph instance makes are governed.View AI clients and agentsAmpAmp's managed settings route its model calls to Argent, attributed to an agent with an owner.View AI clients and agentsWarpA Warp custom inference endpoint pointed at Argent puts the Warp agent's model calls under policy.View AI clients and agentsXcode IntelligenceAdd Argent as a custom provider in Xcode and its coding assistant calls are checked and recorded.View AI clients and agentsFactory DroidA Factory custom model pointed at Argent puts Droid's calls under your policy and spend caps.View AI clients and agentsQwen CodeQwen Code's endpoint and proxy settings route its calls through Argent for checks and signed records.View AI clients and agentsGoogle Antigravity CLIPoint the Antigravity CLI's endpoint at Argent and every call is attributed, checked and recorded.View AI clients and agentsMistral Vibe CLIMistral Vibe's provider setting points at Argent, so its calls land under policy.View AI clients and agentsOpenHandsOpenHands CLI and self-hosted Enterprise send their model calls to your Argent route.View AI clients and agentsClaude Desktop on third-party inferenceClaude Desktop's gateway mode points at Argent, so every chat is attributed, checked and signed.View AI clients and agentsSalesforce AgentforceModels you register through Salesforce's LLM Open Connector can run through Argent.View AI clients and agentsServiceNow Now AssistA Generative AI Controller connection pointed at Argent puts Now Assist model calls under policy.View AI clients and agentsOpenAI SDKsApps on openai-python or openai-node reach Argent with one base URL or a proxy variable.View AI clients and agentsAnthropic SDKsApps on the Anthropic SDKs call Argent by base URL or proxy and every call is governed.View AI clients and agentsAmazon Bedrock SDKsBedrock calls go through Argent, which signs them with a brokered AWS credential your app never holds.View AI clients and agentsAzure OpenAIAzureOpenAI clients point at Argent and every deployment call is attributed, checked and recorded.View AI clients and agentsGoogle Gen AI and Vertex AIGemini API and Vertex AI calls from the Google Gen AI SDK route through Argent for policy and records.View AI clients and agentsLangChainLangChain chat models point at Argent, or use the proxy with no code change, and every call is governed.View AI clients and agentsLangGraphLangGraph agents send model calls through Argent, so each step of a graph is attributed and recorded.View AI clients and agentsLlamaIndexLlamaIndex LLM classes take an Argent base URL, so RAG and agent calls land under policy.View AI clients and agentsCrewAIEach CrewAI agent's model calls route through Argent and are attributed, checked and recorded.View AI clients and agentsOpenAI Agents SDKAgents built on openai-agents send every model call through Argent, by base URL or proxy.View AI clients and agentsClaude Agent SDKAgents built on the Claude Agent SDK get Argent's gateway, proxy and hooks, like Claude Code.View AI clients and agentsn8nSelf-hosted n8n workflows send their OpenAI and Anthropic node calls through Argent.View AI clients and agentsVercel AI SDKAI SDK providers take an Argent base URL, so app model calls are attributed, checked and recorded.View AI clients and agentsMicrosoft Agent FrameworkAgent Framework, Semantic Kernel and AutoGen apps route their model calls through Argent.View AI clients and agentsLiteLLMLiteLLM can call Argent as its upstream, so every model call it routes is governed and signed.View AI clients and agentsOpen WebUIOpen WebUI's model connections point at Argent, so every chat your people send is governed.View AI clients and agentsOllamaArgent fronts your Ollama server: people and apps call Argent, and Ollama answers only Argent.View AI clients and agentsLM StudioRegister LM Studio's local server as an Argent upstream and every call to it is governed and signed.View AI clients and agentsPerplexity APIOpenAI-compatible clients reach Perplexity Sonar through Argent, under your policy and caps.View AI clients and agentsMistral APICalls to Mistral's La Plateforme go through Argent, attributed, redacted, capped and recorded.View AI clients and agentsDeepSeek APICalls to the DeepSeek API go through Argent, where policy decides who may use it and what data goes.View AI clients and agentsxAI Grok APICalls to the xAI Grok API go through Argent, attributed to an owned agent and signed.View AI clients and agentsLibreChatA LibreChat custom endpoint pointed at Argent puts every chat under your policy and records.View MCP GateMCP serversEvery MCP tool call checked against a per-tool permission at the gateway, refused by default, and signed.View MCP registryOfficial MCP RegistryImport servers from the public MCP Registry switched off; an admin approves each one and its tools.View MCP registryDocker MCP CatalogImport servers from Docker's curated MCP Catalog switched off; an admin approves each one and its tools.View MCP registryInternal MCP registryThe live list of servers the MCP Gate governs. Not on the list, no call. Bring your own registry too.View A2A GateA2A agent-to-agentAgent-to-agent calls checked at the gateway: a rule and a permission for every job, refused by default, signed.View DirectoryOktaPulls people, groups and memberships from Okta; a leaver's agents stop on Argent's next check.View DirectoryMicrosoft Entra IDMirrors Entra Agent ID agents and makes each agent's human sponsor its owner in Argent.View GovernanceSailPoint IdentityIQIdentityIQ provisions people and groups over SCIM; in federated mode Argent pushes kills back over SCIM.View GovernanceSailPoint Identity Security CloudIdentity Security Cloud provisions people and groups over SCIM; in federated mode Argent pushes kills back over SCIM.View GovernanceSaviyntSaviynt provisions people and groups over SCIM; in federated mode Argent pushes kills back over SCIM.View GovernancePing IdentityPing Identity provisions people and groups over SCIM; in federated mode Argent pushes kills back over SCIM.View DirectoryAuth0Pulls users and roles from your Auth0 tenant, so every agent answers to an active person.View DirectoryKeycloakPulls users and groups from a Keycloak realm; Keycloak can also be your console sign-in.View DirectoryLDAP and Active DirectoryReads people and groups from Active Directory or any LDAP directory, read-only.View DirectoryPostgreSQL user tablesReads people from an application's own PostgreSQL user table, read-only.View ProvisioningSCIM 2.0A SCIM 2.0 endpoint identity providers and governance tools push into; in federated mode kills go back upstream.View Sign-inOpenID Connect SSOConsole and owner-portal sign-in through your OpenID Connect provider, per tenant, proven with Keycloak and Entra ID.View Sign-inSAML 2.0 SSOThe same console sign-in over SAML 2.0, with signature, audience and destination checks on every assertion.View Sign-inPasskeysPasswordless admin sign-in, plus a fresh passkey for every sensitive change.View Secrets vaultsHashiCorp Vault and OpenBaoYour Vault or OpenBao cluster keeps the AI keys. Argent reads each KV v2 path on use and attaches the key at the gateway.View Secrets vaultsAWS Secrets ManagerKeys stay in AWS Secrets Manager. Argent fetches each by id or ARN on use and attaches it at the gateway.View Secrets vaultsCyberArk ConjurKeys stay in your Conjur appliance. Argent signs in as a host identity and fetches each variable when a request needs it.View Secrets vaultsInfisicalKeys stay in Infisical, cloud or self-hosted. Argent signs in with Universal Auth and reads each secret by name on use.View Built inOpen Policy Agent (built in)The default decision point for every request: Rego rules, default deny, fail closed, editable and testable in the console.View Bring your ownOpen Policy Agent (your server)Route a tenant's decisions to your standalone OPA server over its REST API. Only an explicit allow passes; errors deny.View Bring your ownCerbosYour Cerbos PDP checks each agent request; agent permissions become Cerbos roles. Only an explicit allow passes; errors deny.View Bring your ownAuthZed SpiceDBSpiceDB checks whether each agent holds a permission on what it calls. Only a definite yes passes; errors deny.View Bring your ownTopazYour Topaz authorizer decides every agent request in the open AuthZEN format. Only an explicit yes passes.View RAGOpenFGA for RAGRelationship-based RAG authorization: retrieved documents filtered to what the agent's human owner may read, fail closed.View Data protection enginesPresidio (built in)The built-in engine on every gateway. Finds personal data and secrets in prompts and replies, then redacts, masks, hashes or blocks them.View Data protection enginesONNX injection classifierA small prompt-injection model that runs on plain CPUs inside your deployment. Weights are fetched by you, license-gated and pinned.View Data protection enginesLLM Guard (Protect AI)Your LLM Guard scanners judge every prompt at the gateway, and replies when reply checks are on, under one Argent policy.View Data protection enginesAmazon Bedrock GuardrailsYour Bedrock guardrail checks every prompt at the gateway through ApplyGuardrail, and replies when reply checks are on, for any model.View Cloud and deploy targetsAmazon Web ServicesArgent runs on EKS, governs Bedrock, and scans your AWS account read-only to show whether AWS blocks ways around it.View Cloud and deploy targetsMicrosoft AzureArgent runs on AKS, governs Azure OpenAI and Foundry, and scans Azure read-only to show whether Azure blocks ways around it.View Cloud and deploy targetsGoogle CloudArgent runs on GKE, governs Google Gemini, and scans your project read-only to show whether Google blocks ways around it.View Cloud and deploy targetsKubernetes and HelmOne Helm chart for EKS, AKS, GKE, k3s and RKE2. Signed images, two doors to the network, in-cluster or managed data.View Cloud and deploy targetsDocker ComposeOne signed bundle for one Linux server: host checks, signature checks and per-install secrets, with no repository to clone.View Cloud and deploy targetsAir-gapped installOne signed bundle carried into a sealed network. Every signature, checksum and image is checked on your side before it loads.View Managed-device rolloutJamfOne Argent script in a Jamf Pro policy brings the command-line AI tools on every managed Mac into view. Public certificate only, reversible.View Managed-device rolloutKandji (now Iru)The same Mac script as a Kandji (now Iru) Custom Script on a Blueprint. Runs as root, safe to repeat, reversible.View Managed-device rolloutMicrosoft IntuneOne PowerShell script as an Intune remediation, run as SYSTEM, brings the command-line AI tools on every managed Windows PC into view.View Ticketing and approvalsServiceNowAI budget and MCP access requests open ServiceNow tickets; the decision lands on the ticket, signed.View Ticketing and approvalsJira Service ManagementTickets for AI budget and MCP requests; a native Approve or Decline is applied under Argent's checks.View Ticketing and approvalsJira (work tracking)AI spend booked to the Jira story it served, capped, and written back to Jira, signed.View Ticketing and approvalsGitHub (work tracking)AI spend booked to the GitHub issue it served, capped, and written back to GitHub, signed.View Ticketing and approvalsGitLab (work tracking)AI spend booked to the GitLab issue it served, capped, and written back to GitLab, signed.View Ticketing and approvalsLinear (work tracking)AI spend booked to the Linear issue it served, capped, and written back to Linear, signed.View Ticketing and approvalsNotion (work tracking)AI spend booked to the Notion page it served, capped, and written back to Notion, signed.View Notifications and alertingSlackBudget, skill approval and note mention alerts posted to a Slack channel through an incoming webhook.View Notifications and alertingPagerDutyBudget and skill approval alerts raised as PagerDuty incidents through Events API v2, one incident per event.View Notifications and alertingEmail (SMTP)Alerts, audit-break notices and budget owner notices by email through your own relay, STARTTLS required.View Notifications and alertingSigned webhookAlerts as JSON to any HTTPS endpoint you run, optionally signed with HMAC-SHA256 so every delivery can be verified.View Observability stackPrometheus, Alertmanager and GrafanaAbout 160 metric series, alert rules with unit tests and five Grafana dashboards, bundled with the Compose stack and Helm chart.View Observability stackOpenTelemetry Collector, Tempo and LokiTraces of every request across gateway, controls and control plane over OTLP, logs in Loki; a trace ID matches its signed audit row.View Log, SIEM and proxy importsSquid proxy logsFinds which AI services people reach from a copy of your Squid access log. Metadata only, with silence detection.View Log, SIEM and proxy importsGeneric proxy NDJSONThe same discovery from any proxy or log pipeline that writes the documented ten-field NDJSON schema.View Log, SIEM and proxy importsOpenTelemetry (OTLP) from coding toolsUsage telemetry from Claude Code, Codex CLI and Gemini CLI, rolled up per agent into signed records, content dropped.View