# AxioRank

> AxioRank is the security gateway for AI agents. Before any agent touches your MCP servers, APIs, and databases, it routes the call through a gateway that gives the agent a short-lived identity, enforces your policy, scores the risk, and writes a redacted audit trail in under 100ms. It also verifies the AI agents that visit your site.

## Product

- [Home](https://axiorank.com/): The security gateway for AI agents, in both directions.
- [Features](https://axiorank.com/features): The full security gateway for AI agents: identity, policy, inspection, audit, and proof.
- [Agent Identity](https://axiorank.com/features/identity): Short-lived, verifiable identity for every agent.
- [AI Gateway](https://axiorank.com/features/ai-gateway): An OpenAI-compatible gateway that routes, fails over, guards by default, and signs a verifiable receipt for every response. Includes a live playground.
- [Model catalog](https://axiorank.com/models): Every model the AI Gateway reaches by name, with context windows, prices per 1M tokens, and copy-paste calls.
- [MCP Gateway](https://axiorank.com/features/mcp-gateway): Govern every MCP tool call through one gateway.
- [Policy Engine](https://axiorank.com/features/policy-engine): Resource, agent, and ML predicates as one-line rules.
- [Content Inspection](https://axiorank.com/features/content-inspection): Detect secrets, PII, destructive operations, prompt injection, and egress.
- [Automated Response](https://axiorank.com/features/automated-response): Quarantine, revoke, alert, and notify automatically.
- [Threat Intelligence](https://axiorank.com/features/threat-intelligence): Cross-tenant threat intel that enriches every verdict.
- [Provable Security](https://axiorank.com/features/provable-security): Information-flow control and a verifiable transparency log.
- [Agent Runs](https://axiorank.com/features/agents): Every agent run as one trace: steps, decisions, cost, latency, and signed proof.
- [Agent observability](https://axiorank.com/agent-observability): What agent observability is and how AxioRank records it from the security gateway, with verdicts and proof on every step.
- [Pricing](https://axiorank.com/pricing): Start free. Scale to Enterprise.
- [Protocol coverage](https://axiorank.com/protocols): Every agent-interop protocol AxioRank governs, across six planes.
- [Benchmarks](https://axiorank.com/benchmarks): How AxioRank's enforcement is measured.
- [Detection benchmark](https://axiorank.com/benchmarks/detection): AxioRank's attack catch-rate next to open-source guardrails and baselines on public and red-team corpora, at a fixed low false-positive budget.
- [Attack library](https://axiorank.com/attacks): The red-team corpus the engine is tested against, with how each attack is caught.
- [Developers](https://axiorank.com/developers): SDKs, the OpenAPI spec, webhooks, and framework adapters.
- [Integrations](https://axiorank.com/integrations): Drop-in adapters for OpenAI Agents, LangChain, Vercel AI SDK, CrewAI, and more.
- [AGS-1 specification](https://axiorank.com/spec): The open Agent Security Gateway Specification: the five gates every agent action passes through.
- [Protected by AxioRank](https://axiorank.com/protected): The badge that shows an agent routes its tool calls through the AxioRank security gateway.
- [Calculators](https://axiorank.com/calculators): Interactive calculators that estimate AI agent breach cost, risk exposure, and the ROI of governance.
- [Compare](https://axiorank.com/compare): How AxioRank compares to alternatives.
- [Blog](https://axiorank.com/blog): Field notes on securing AI agents, prompt injection, and MCP verification.
- [Trust registry](https://axiorank.com/trust): The public registry of verified agent hosts.
- [Verify a site](https://axiorank.com/verify): Verify an agent host's posture.
- [Security](https://axiorank.com/security): AxioRank's own security posture.
- [Documentation](https://axiorank.com/docs): Guides, API reference, and integrations.

## SDKs & CLIs

- [@axiorank/sdk (npm)](https://www.npmjs.com/package/@axiorank/sdk): TypeScript SDK to inspect or govern any tool call, with adapters for OpenAI Agents, LangChain, the Vercel AI SDK, LlamaIndex, Mastra, Anthropic, and Gemini.
- [axiorank (PyPI)](https://pypi.org/project/axiorank/): Python SDK with adapters for LangChain, LlamaIndex, CrewAI, Pydantic AI, OpenAI Agents, LiteLLM, and more.
- [@axiorank/mcpaudit (npm)](https://www.npmjs.com/package/@axiorank/mcpaudit): CLI that scans MCP servers for prompt injection, tool poisoning, and leaked secrets. Run npx @axiorank/mcpaudit scan --share for a shareable scorecard.
- [@axiorank/gateway (npm)](https://www.npmjs.com/package/@axiorank/gateway): npx-runnable OpenAI-compatible AI gateway with 34 provider presets, guardrails on by default, and a signed, offline-verifiable receipt on every response. Open source (MIT) at https://github.com/AxioRank/gateway.
- [@axiorank/mcp-gateway (npm)](https://www.npmjs.com/package/@axiorank/mcp-gateway): Transparent MCP proxy that routes any MCP server's tool calls through AxioRank governance.
- [@axiorank/audit-verify (npm)](https://www.npmjs.com/package/@axiorank/audit-verify): Zero dependency CLI and library to verify audit receipts and Agent Action Seals offline.
- [@axiorank/coding-guard (npm)](https://www.npmjs.com/package/@axiorank/coding-guard): Hook that governs coding agents (Claude Code, Cursor, Codex CLI, and Copilot CLI) locally. Install with npx @axiorank/coding-guard init.
- [Remote MCP server](https://app.axiorank.com/api/mcp-server/mcp): Add AxioRank as an MCP server to manage policies and query the audit trail from your own agent. Listed in the official MCP registry as com.axiorank/axiorank.

## Documentation

- [Agent Runs (traces, sessions, metadata)](https://axiorank.com/docs/agent-runs): Group an agent's tool calls and model turns into one run, group runs into sessions, tag calls with your own metadata, and read it all back with per-step cost and latency.
- [AGS-1 Conformance](https://axiorank.com/docs/ags-1-conformance): How to conform to the Agent Security Gateway Specification, how AxioRank satisfies each of the five gates, how to declare a scoped conformance claim, and how to publish the descriptor and badge.
- [AGS-1 and the OWASP Top 10](https://axiorank.com/docs/ags-1-owasp): How the five AGS-1 gates map to the OWASP Top 10 for Agentic Applications (2026) and the OWASP Top 10 for LLM Applications (2025). An informative crosswalk.
- [AGS-1 Specification](https://axiorank.com/docs/ags-1): The Agent Security Gateway Specification. The five gates every agent action passes through before it reaches an MCP server, an API, or a database. RFC 2119.
- [AI Gateway (drop-in proxy)](https://axiorank.com/docs/ai-gateway): Govern every model call across your org by changing one base URL. No application code changes.
- [API Reference](https://axiorank.com/docs/api-reference): Interactive reference for the AxioRank gateway API, generated from the live OpenAPI 3.1 spec.
- [Approvals](https://axiorank.com/docs/approvals): Hold a risky call for a human decision, then approve or deny from the dashboard, a signed email link, or Slack.
- [Assurance center](https://axiorank.com/docs/assurance): One health view for the parts of AxioRank that run in the background, your integrations, the ML lane, data residency, and the async job queue, so a silent degradation never hides.
- [Audit log export](https://axiorank.com/docs/audit-export): Pull redacted audit rows as NDJSON into your SIEM, or download CSV from the Logs page in one click.
- [Audit integrity](https://axiorank.com/docs/audit-integrity): Prove the audit log hasn't been altered with a sealed, tamper-evident hash chain.
- [Browser agent security](https://axiorank.com/docs/browser-agents): Govern AI agents that act inside the browser with a Chrome extension that scans pages for prompt injection, blocks secret and PII exfiltration through forms, network requests, and the clipboard, and mints a signed, offline-verifiable Browser Session Seal.
- [Agent budgets](https://axiorank.com/docs/budgets): Cap what an agent can spend and do, per hour, day, or month, with the same policy engine that governs every call.
- [Bring your own key (BYOK)](https://axiorank.com/docs/byok): Wrap your broker secrets with your own AWS KMS key so revoking the grant provably locks AxioRank out of your credentials.
- [Changelog](https://axiorank.com/docs/changelog): Developer-facing changes to the AxioRank API, SDKs, and webhooks.
- [Claims and verification](https://axiorank.com/docs/claims): Every "first" or "only" claim AxioRank makes, stated precisely, with how to verify it yourself.
- [Coding agent security](https://axiorank.com/docs/coding-agents): Govern Claude Code, Cursor, Codex, and Copilot CLI with a local hook that blocks dangerous tool calls and mints a signed, offline-verifiable Coding Session Seal.
- [Config as code](https://axiorank.com/docs/config-as-code): Export, review, and import your workspace governance as a versioned JSON document.
- [Content-inspection engine](https://axiorank.com/docs/content-inspection): Payload-aware detection of secrets, PII, destructive actions, injection, malware, and crypto/financial identifiers, on both tool arguments and tool outputs.
- [AI Governance Copilot](https://axiorank.com/docs/copilot): Ask AxioRank questions about your own governance data in plain language. A read-only analyst grounded in your agents, policies, incidents, logs, and spend.
- [Data residency](https://axiorank.com/docs/data-residency): Where AxioRank stores your data, exactly which sub-processor holds what, and the controls EU and UK teams use to keep data under their governance.
- [Detection benchmark methodology](https://axiorank.com/docs/detection-benchmark): How the competitive detection benchmark works. The normalized case schema, the adapter contract, the tools we run and the ones we cannot, the corpora and how benign controls are built, the catch-rate-at-a-false-positive-budget metric, and the pre-registered gate. Everything you need to reproduce every number.
- [Detection intelligence](https://axiorank.com/docs/detection-intelligence): How a call becomes a verdict beyond the deterministic signals, with ML assessment, a semantic judge, taint provenance, and kill-chain correlation.
- [Emergency lockdown](https://axiorank.com/docs/enforcement-floor): The workspace-wide circuit breaker. Freeze every governed call in one action when something goes wrong, then release case by case.
- [EU AI Act evidence pack](https://axiorank.com/docs/eu-ai-act): Offline-verifiable evidence supporting Article 12 (logging) and Article 14 (human oversight).
- [Compliance evidence bundle](https://axiorank.com/docs/evidence-bundle): One zip of offline-verifiable evidence for SOC 2, ISO 27001, HIPAA, and more, assembled live from your control plane.
- [Gateway API](https://axiorank.com/docs/gateway-api): The raw HTTP contract behind every AxioRank SDK, callable from any language.
- [Open-source gateway (npx)](https://axiorank.com/docs/gateway-oss): Run the AxioRank AI Gateway on your own machine. One command, guardrails on by default, and a signed receipt on every response. MIT licensed, zero runtime dependencies.
- [Getting started](https://axiorank.com/docs/getting-started): Route your first agent tool call through the AxioRank gateway.
- [GitHub Action](https://axiorank.com/docs/github-action): Fail the build when a repo references an untrusted MCP server or a fixture your policies should deny.
- [GitOps](https://axiorank.com/docs/gitops): Run your workspace governance like infrastructure, with the config file in git, a drift gate on pull requests, and an import on merge.
- [Inbound surfaces](https://axiorank.com/docs/inbound-surfaces): Verify the AI agents that reach into the surfaces you operate, a website, an HTTP API, an MCP server, an A2A agent, or a webhook, with one verify endpoint and a per-kind challenge.
- [Introduction](https://axiorank.com/docs): Zero-Trust security and identity for AI agents. One control plane that governs every tool call.
- [MCP gateway](https://axiorank.com/docs/mcp-gateway): Govern an MCP server's tool calls with no changes to your agent's code.
- [AxioRank MCP server](https://axiorank.com/docs/mcp-server): Call AxioRank governance and control-plane capabilities as tools from any MCP client.
- [Organizations](https://axiorank.com/docs/organizations): Group your workspaces under one organization for consolidated billing, a shared plan, and config templates you push across teams.
- [Policies](https://axiorank.com/docs/policies): Turn a risk score into an enforceable verdict per agent, tool, and request attribute.
- [Privacy & data rights](https://axiorank.com/docs/privacy): Jurisdiction-aware data-subject requests for access, deletion, correction, and opt-out.
- [Protocol adapters](https://axiorank.com/docs/protocol-adapters): Apply the gateway across the agent-interop surface, spanning identity, messaging, authorization, content, and commerce.
- [Python SDK](https://axiorank.com/docs/quickstart-python): Route AI agent tool calls through AxioRank from Python (sync or async).
- [TypeScript SDK](https://axiorank.com/docs/quickstart-typescript): Route outbound tool calls and verify inbound agents from TypeScript.
- [Response engine](https://axiorank.com/docs/response-engine): Act automatically on risk by quarantining an agent, revoking a key, or raising an alert when a rule fires.
- [SCIM provisioning](https://axiorank.com/docs/scim): Let Okta or Entra ID create, deactivate, and regroup AxioRank members automatically.
- [Secrets broker](https://axiorank.com/docs/secrets-broker): Hold upstream credentials in a vault and inject them at the proxy so the agent never sees the secret.
- [Shadow AI discovery](https://axiorank.com/docs/shadow-ai-discovery): Find the AI agents, models, and MCP servers your org already uses but does not govern, from egress logs, MCP scans, identity events, and cloud audit trails.
- [Spend governance](https://axiorank.com/docs/spend-governance): See what every agent and model is costing you over time, set a workspace budget, and act on cost from the same gateway that already governs risk.
- [Versioning & stability](https://axiorank.com/docs/stability): What's stable, what counts as a breaking change, and how we ship them.
- [Verify our log yourself](https://axiorank.com/docs/verify-our-log): Pin our public key, pull the signed checkpoints, and verify receipts offline. Trust nothing of ours.
- [Voice agents (Realtime API)](https://axiorank.com/docs/voice-agents): Govern the tool calls and output of an OpenAI Realtime voice agent with the same policy, risk scoring, and audit trail as any other agent.
- [Webhooks](https://axiorank.com/docs/webhooks): Subscribe your services to AxioRank events with signed, retried, replayable delivery.
- [Workspace security](https://axiorank.com/docs/workspace-security): Roles, SAML SSO, and workspace MFA, plus how the API reports each policy failure.
- [Google ADK](https://axiorank.com/docs/integrations/adk): Guard a Google ADK tool or plain callable with AxioRank before it runs.
- [Adobe Commerce](https://axiorank.com/docs/integrations/adobe-commerce): Verify the AI agents that reach your Adobe Commerce and Magento Open Source REST API, GraphQL, login, and checkout with the AxioRank Agent Verification extension. A thin client of the inbound verify endpoint, with monitor and enforce modes.
- [Anthropic Messages API](https://axiorank.com/docs/integrations/anthropic): Guard the tool_use loop you run yourself with the Anthropic SDK, so every dispatch is scored before it executes.
- [Atlassian (Jira and Confluence)](https://axiorank.com/docs/integrations/atlassian): Verify and govern the AI agents in your Atlassian tenant, including Rovo, with the AxioRank Agent Security Forge app. A thin client of the inbound verify endpoint that guards Rovo actions, verifies external agents cryptographically, and records agent provenance.
- [AutoGen / AG2](https://axiorank.com/docs/integrations/autogen): Guard an AutoGen or AG2 tool function with AxioRank before registering it.
- [Backdrop CMS](https://axiorank.com/docs/integrations/backdrop): Verify the AI agents that reach your Backdrop CMS login, admin, search, AJAX, and API endpoints with the AxioRank Agent Verification module. A thin client of the inbound verify endpoint, with monitor and enforce modes and in-admin rule management.
- [Cloudflare Agents](https://axiorank.com/docs/integrations/cloudflare-agents): Govern a Cloudflare Agent's tools with AxioRank. The Agents SDK uses Vercel AI SDK tools, so the AxioRank Vercel adapter guards them with no rewrite.
- [CrewAI](https://axiorank.com/docs/integrations/crewai): Guard a CrewAI tool or an entire toolbelt with AxioRank before it runs.
- [Datadog](https://axiorank.com/docs/integrations/datadog): Stream the AxioRank governed audit log into Datadog and install the AxioRank tile for a prebuilt log pipeline, three dashboards, and five monitors. Watch agent governance next to the rest of your stack.
- [Drupal](https://axiorank.com/docs/integrations/drupal): Verify the AI agents that reach your Drupal REST, JSON:API, admin, and dynamic endpoints with the AxioRank Agent Verification module. A thin client of the inbound verify endpoint, with monitor and enforce modes.
- [Gemini Enterprise Agent Platform](https://axiorank.com/docs/integrations/gemini-enterprise): Govern agents built on Google's Gemini Enterprise Agent Platform. One ADK plugin for the whole runtime, plus model-call guarding, card verification, and shadow-agent discovery.
- [Google Gemini](https://axiorank.com/docs/integrations/gemini): Guard the function-calling loop you run with the Gemini SDK, so every dispatch is scored before it executes.
- [HubSpot](https://axiorank.com/docs/integrations/hubspot): Install AxioRank from the HubSpot Marketplace to govern the AI agents that reach your CRM through HubSpot's remote MCP server. A CRM card shows that protection on every contact, company, and deal.
- [Framework integrations](https://axiorank.com/docs/integrations): Drop-in adapters that put every tool your agent runs behind the AxioRank gateway, no rewrite.
- [LangChain · LangGraph](https://axiorank.com/docs/integrations/langchain): Put every tool a LangChain agent runs behind the AxioRank gateway, in TypeScript or Python.
- [LangGraph](https://axiorank.com/docs/integrations/langgraph): Govern every tool a LangGraph agent runs through the AxioRank gateway.
- [LiteLLM proxy](https://axiorank.com/docs/integrations/litellm): Govern prompts, completions, and tool calls behind a LiteLLM proxy with one AxioRank guardrail.
- [LlamaIndex](https://axiorank.com/docs/integrations/llamaindex): Guard LlamaIndex tools with AxioRank before they run, in TypeScript or Python.
- [Mastra](https://axiorank.com/docs/integrations/mastra): Put every tool a Mastra agent runs behind the AxioRank gateway.
- [n8n](https://axiorank.com/docs/integrations/n8n): Drive AxioRank from an n8n workflow with the community node. Trigger a workflow on a governance event, and act back to approve a held call, quarantine an agent, or respond to an incident.
- [OpenAI Agents](https://axiorank.com/docs/integrations/openai-agents): Guard the tools an OpenAI Agents SDK agent runs, in TypeScript or Python.
- [OpenRouter](https://axiorank.com/docs/integrations/openrouter): Govern agents that route models through OpenRouter with AxioRank. OpenRouter speaks the OpenAI API, so the AxioRank OpenAI adapter guards every tool call with no OpenRouter-specific package.
- [PagerDuty](https://axiorank.com/docs/integrations/pagerduty): Page your on-call for AI agent incidents in PagerDuty. One PagerDuty incident per correlated AxioRank incident, with trigger, acknowledge, and resolve kept in sync both ways.
- [Pydantic AI](https://axiorank.com/docs/integrations/pydantic-ai): Guard a Pydantic AI tool with AxioRank before it runs.
- [Shopify](https://axiorank.com/docs/integrations/shopify): Verify the AI agents that browse and buy on your Shopify store with the AxioRank Agent Verification app. A thin client of the inbound verify endpoint, with monitor and enforce modes and order-level enforcement.
- [Slack](https://axiorank.com/docs/integrations/slack): Add AxioRank to Slack with one click. Route high-risk AI-agent alerts to a channel, approve or deny held tool calls right in Slack, and run /axiorank for what needs attention.
- [smolagents](https://axiorank.com/docs/integrations/smolagents): Wrap any smolagents Tool so AxioRank scores every execution, with the schema untouched.
- [Strands Agents](https://axiorank.com/docs/integrations/strands): Govern a Strands (AWS) agent's tools with AxioRank. Wrap each @tool so every call is scored before it runs, with no rewrite.
- [Microsoft Teams](https://axiorank.com/docs/integrations/teams): Bring AxioRank into Microsoft Teams. Route high-risk AI-agent alerts to a chat or channel and approve or deny held tool calls right in Teams, with Bot Framework verified callbacks and a tamper-evident audit trail.
- [TYPO3](https://axiorank.com/docs/integrations/typo3): Verify the AI agents that log in, call JSON endpoints, search, and submit forms on your TYPO3 site with the AxioRank Agent Verification extension. A native TYPO3 v13 extension and a thin client of the inbound verify endpoint that blocks unverified agents in real time.
- [Vercel AI SDK](https://axiorank.com/docs/integrations/vercel): Put every tool a Vercel AI SDK agent can call behind the AxioRank gateway.
- [WordPress](https://axiorank.com/docs/integrations/wordpress): Verify the AI agents that reach your WordPress REST API, admin, and dynamic endpoints with the AxioRank Agent Verification plugin. A thin client of the inbound verify endpoint, with monitor and enforce modes.
- [Zapier](https://axiorank.com/docs/integrations/zapier): Connect AxioRank to Zapier so a governance event can trigger a Zap, and a Zap can approve a held call, quarantine an agent, or respond to an incident. No code.

## Blog

- [How to block fake ChatGPT bots in WordPress](https://axiorank.com/blog/block-fake-chatgpt-bots-wordpress): Anyone can set a User-Agent to ChatGPT, so a name-based block list cannot catch impostors. Here is how to verify the real OpenAI crawlers on your WordPress site and block the fakes, without writing code.
- [Prompt injection vs tool injection: two doors into your agent](https://axiorank.com/blog/prompt-injection-vs-tool-injection): Prompt injection poisons what the model reads. Tool injection poisons the tools it can call. Here is how to tell them apart, why the difference changes your defense, and the one place both attacks can be stopped.
- [Govern the agent writing your code: AxioRank for Claude Code, Cursor, Codex, and Copilot](https://axiorank.com/blog/govern-ai-coding-agents): AI coding agents run commands, edit files, and call tools on your machine before you can review them. AxioRank now governs that loop with one hook, blocking dangerous actions locally and minting a signed, offline-verifiable proof that the session was governed.
- [Attack explained: the credential exfiltration kill chain](https://axiorank.com/blog/aws-key-exfiltration-kill-chain): A deep walkthrough of the most common agent attack, reading a secret and sending it somewhere it should not go. See it run, see why single-call checks miss it, and see exactly how value-level taint tracking catches it.
- [Verify an MCP server before you connect to it](https://axiorank.com/blog/verify-mcp-server-before-connecting): An MCP server you connect to can read your context and propose tool calls. Check its card, signature, and supply-chain risk first, the same way you would vet a dependency.
- [Prompt injection is a tool-call problem](https://axiorank.com/blog/prompt-injection-in-tool-calls): Prompt injection only matters when it makes an agent do something. Catching it at the tool call, not the prompt, is what actually stops the harm.
- [What Zero-Trust means for AI agents](https://axiorank.com/blog/zero-trust-for-ai-agents): Agents act on your behalf with real credentials and real tools. Zero-Trust for agents means never trusting a tool call by default, and verifying every one before it runs.

## Machine-readable interop

- [Protocol coverage (JSON)](https://axiorank.com/api/protocols): The protocol matrix as JSON.
- [OpenAPI 3.1 spec](https://axiorank.com/api/openapi.json): The full REST API description.
- [MCP server card](https://axiorank.com/.well-known/mcp.json): AxioRank's MCP server card and tools.
- [A2A agent card](https://axiorank.com/.well-known/agent.json): AxioRank's signed A2A agent card.
- [Agent Skills](https://axiorank.com/.well-known/agent-skills.json): AxioRank's declared agent skills.
- [WebMCP manifest](https://axiorank.com/.well-known/webmcp.json): Page tools AxioRank exposes to visiting agents.
- [OAuth authorization server](https://axiorank.com/.well-known/oauth-authorization-server): OAuth 2.1 discovery metadata.