{ "meta": { "name": "Agent Index — AI agent security and governance", "version": "0.1.3", "generated": "2026-10-30", "license": "CC-BY-4.0", "citation": "Agent Index by Heritage Services (https://theagentindex.io). Agent security dataset version 0.1.3.", "segment": "agent-security", "index_min": 5, "organization": { "name": "Heritage Services", "legal_name": "Heritage Services", "url": "https://theagentindex.io", "email": "sales@heritageservices.online", "gstin": "27ACQPA3590B1ZS", "udyam": "UDYAM-MH-33-0712021" } }, "categories": [ { "slug": "agent-identity", "name": "Agent identity", "description": "Who an AI agent is: authentication, authorization, and workload identity for agents and their tools." }, { "slug": "agent-governance", "name": "Agent governance", "description": "Policies, approvals, and audit trails for what AI agents may do and who reviews them." }, { "slug": "agent-runtime-security", "name": "Agent runtime security", "description": "Runtime controls, isolation, and monitoring while an AI agent is running." }, { "slug": "agent-secrets", "name": "Agent secrets", "description": "Secrets, credentials, and access material used by AI agents and their tool calls." } ], "agents": [ { "slug": "descope-agentic-identity-hub", "name": "Descope Agentic Identity Hub", "vendor": "Descope", "hq_country": "United States", "hq_region": "north-america", "url": "https://www.descope.com/use-cases/ai", "segment": "agent-security", "category": "agent-identity", "categories": [ "agent-identity", "agent-governance" ], "summary": "Identity provider for AI agents and MCP servers: OAuth-based auth, consent, scoped short-lived agent credentials, access policies and a directory to monitor and revoke agents.", "description": "The Descope Agentic Identity Hub is a dedicated identity provider for AI agents. Organizations use it to secure their MCP servers with OAuth and enterprise-managed authorization, issue short-lived scoped credentials for their AI agents, and treat AI agents as dedicated, first-class identities.", "pricing": "", "pricing_model": "published-tiers", "frameworks": [ "soc2", "iso27001", "hipaa", "fedramp", "dora" ], "frameworks_claimed": [], "claimed": true, "claimed_at": "2026-10-30", "tasks": [ "agent-authentication", "agent-authorization", "mcp-access-control", "agent-credential-management", "agent-inventory", "audit-logging" ], "india_relevance": "", "featured": false, "source_url": "https://www.descope.com/use-cases/ai", "sources": [ "https://www.descope.com/use-cases/ai", "https://docs.descope.com/", "https://www.descope.com/pricing", "https://www.descope.com/security-compliance", "https://www.descope.com/blog/post/agentic-identity-hub-3", "https://www.descope.com/blog/post/dora-auth-mfa", "https://docs.descope.com/fedramp", "https://trust.descope.com/", "https://www.descope.com/privacy", "https://www.descope.com/favicon.ico" ], "last_checked": "2026-10-30", "added": "2026-10-29" }, { "slug": "saviynt-zuma", "name": "Saviynt Zuma (AI identity security)", "vendor": "Saviynt", "hq_country": "United States", "hq_region": "north-america", "url": "https://saviynt.com/products/zuma", "segment": "agent-security", "category": "agent-identity", "categories": [ "agent-identity", "agent-governance" ], "summary": "Enterprise AI identity security platform that discovers every AI agent, MCP server and non-human identity, assigns owners, governs their access and blocks anomalous agent actions.", "description": "Saviynt describes Zuma as a control plane for AI agents and non-human identities: it automatically discovers agents, tools, MCP servers and NHIs, keeps one registry with risk insights and named owners, traces the path an agent takes to critical apps and data, and keeps an audit record of agent actions. It evaluates agent intent, context and risk per task and can block anomalous actions, and it distinguishes an agent acting on its own from one acting for a person. Saviynt's page also announces Zuma for Apps for governing agent actions inside business applications in real time, in future tense, so that module is not claimed here.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "agent-discovery", "agent-inventory", "agent-access-governance", "agent-lifecycle-management", "audit-logging" ], "india_relevance": "", "featured": false, "source_url": "https://saviynt.com/products/zuma", "sources": [ "https://saviynt.com/products/zuma", "https://saviynt.com/", "https://saviynt.com/contact-us", "https://saviynt.com/hubfs/raw_assets/saviynt-theme-26/406/js_client_assets/assets/saviynt-logo-DWEmNKg8.svg" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "silverfort-ai-agent-security", "name": "Silverfort AI Agent Security", "vendor": "Silverfort", "hq_country": "United States", "hq_region": "north-america", "url": "https://www.silverfort.com/platform/ai-agent-security/", "segment": "agent-security", "category": "agent-identity", "categories": [ "agent-identity", "agent-runtime-security" ], "summary": "Discovers AI agents across identity providers, cloud and SaaS, enforces least-privilege access at runtime and ties each agent action to a responsible human.", "description": "Silverfort's AI Agent Security continuously discovers sanctioned and rogue AI agents across IdPs, cloud platforms and SaaS apps, assesses their risk and enforces least privilege. Through integrations with Google Cloud Agent Gateway, Microsoft Copilot Studio and its own MCP Gateway it evaluates authorization scopes on each agent call before execution to block privilege escalation, lateral movement or unauthorized access. It maps agent actions to responsible humans and can require approval before agents run sensitive tool actions.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "agent-discovery", "agent-authorization", "least-privilege-enforcement", "mcp-access-control", "human-in-the-loop-approval" ], "india_relevance": "", "featured": false, "source_url": "https://www.silverfort.com/platform/ai-agent-security/", "sources": [ "https://www.silverfort.com/platform/ai-agent-security/", "https://www.silverfort.com/use-cases/secure-ai-agents/", "https://www.silverfort.com/contact/", "https://www.silverfort.com/wp-content/uploads/2025/02/cropped-Silverfort-Symbol-Pos-RGB.png?w=192" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "straiker-defend-ai", "name": "Straiker Defend AI", "vendor": "Straiker", "hq_country": "United States", "hq_region": "north-america", "url": "https://www.straiker.ai/products/defend-ai", "segment": "agent-security", "category": "agent-runtime-security", "categories": [ "agent-runtime-security" ], "summary": "Runtime security for AI agents that inspects prompts, reasoning steps and tool calls to block prompt injection, data exfiltration, agent manipulation and malicious MCP connections.", "description": "Defend AI covers coding agents (Cursor, Claude Code, GitHub Copilot), productivity agents (Microsoft Copilot, ChatGPT Enterprise) and custom agents on AWS Bedrock and Azure AI Foundry. Straiker says its engine is trained on real-world agent traces and inspects every prompt, reasoning step and tool call in real time, blocks destructive actions and malicious MCP connections, and identifies vulnerable MCP servers using its MCP Threat Database. It also traces interactions across the user, model, tool chain and agent-to-agent calls for root-cause analysis.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "prompt-injection-protection", "agent-runtime-monitoring", "mcp-security", "data-exfiltration-prevention", "agent-tracing" ], "india_relevance": "", "featured": false, "source_url": "https://www.straiker.ai/products/defend-ai", "sources": [ "https://www.straiker.ai/products/defend-ai", "https://www.straiker.ai/about", "https://cdn.prod.website-files.com/67e35850d44d18d58bb2404c/68af498d6c908abd19e5d320_straiker-logo-noir.svg" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "operant-agent-protector", "name": "Operant Agent Protector and MCP Gateway", "vendor": "Operant AI", "hq_country": "United States", "hq_region": "north-america", "url": "https://www.operant.ai/platform/agent-protector", "segment": "agent-security", "category": "agent-runtime-security", "categories": [ "agent-runtime-security", "agent-governance" ], "summary": "Inventories AI agents, traces their prompts, tool calls and memory access, and blocks out-of-scope or exfiltrating agent actions inline, with an MCP gateway for the MCP stack.", "description": "Operant's Agent Protector catalogues managed and unmanaged agents and their identities across cloud, SaaS and developer tools, traces activity from prompt through tool calls to memory stores, and scores agent supply-chain risk. When an agent reaches outside its established patterns, escalates privileges or shows signs of exfiltration, it blocks the action inline; Agent ScopeGuard sets the data, APIs and workflows each agent may use. The Operant MCP Gateway adds discovery, detection and defence across local MCP tools and remote agents on Kubernetes, AWS Bedrock, Azure and Google Vertex AI.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "agent-discovery", "agent-runtime-monitoring", "inline-blocking", "mcp-security", "agent-scope-control" ], "india_relevance": "", "featured": false, "source_url": "https://www.operant.ai/platform/agent-protector", "sources": [ "https://www.operant.ai/platform/agent-protector", "https://www.operant.ai/solutions/mcp-gateway", "https://www.operant.ai/company/careers", "https://cdn.prod.website-files.com/679016dc377b83ff1bad1fd4/68f043abdcee353e44c67db9_67906ef7591eaebab1f85dfc_Operant%20-%20Black-p-1080.svg" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "palo-alto-networks-prisma-airs", "name": "Prisma AIRS (AI runtime security)", "vendor": "Palo Alto Networks", "hq_country": "United States", "hq_region": "north-america", "url": "https://www.paloaltonetworks.com/ai-security/prisma-airs", "segment": "agent-security", "category": "agent-runtime-security", "categories": [ "agent-runtime-security", "agent-governance" ], "summary": "AI security platform that discovers, red-teams, protects and governs AI apps, agents and models, with runtime defences against prompt injection, memory manipulation and tool misuse.", "description": "Palo Alto Networks positions Prisma AIRS as securing AI agents, apps, models and data from development to deployment. It includes AI red teaming against agents and applications, runtime protection against prompt injection, model tampering and agentic threats, dedicated AI Agent Security against memory manipulation and tool misuse, and an AI gateway (GA announced July 2026) for governing agent traffic. Palo Alto says agent identity security is now natively available with Prisma AIRS.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "ai-red-teaming", "agent-runtime-monitoring", "prompt-injection-protection", "ai-gateway", "ai-posture-management" ], "india_relevance": "", "featured": false, "source_url": "https://www.paloaltonetworks.com/ai-security/prisma-airs", "sources": [ "https://www.paloaltonetworks.com/ai-security/prisma-airs", "https://www.paloaltonetworks.com/blog/2026/07/announcing-general-availability-of-prisma-airs-ai-gateway/", "https://www.paloaltonetworks.com/about-us/locations", "https://www.paloaltonetworks.com/etc/clientlibs/pan/img/favicons2020/apple-touch-icon-76x76.png" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "palo-alto-networks-idira-secure-ai-agents", "name": "Idira Secure AI Agents (agentic identity security)", "vendor": "Palo Alto Networks", "hq_country": "United States", "hq_region": "north-america", "url": "https://www.paloaltonetworks.com/idira/agentic", "segment": "agent-security", "category": "agent-identity", "categories": [ "agent-identity", "agent-secrets", "agent-governance" ], "summary": "Finds AI agents across SaaS, cloud and developer environments and brokers their identities so each agent gets only task-scoped, time-limited privilege, with logs of what it did and for whom.", "description": "Idira Secure AI Agents scans SaaS, cloud and developer environments for active agents and enriches them with owner, purpose, status and permission context. A dedicated agent identity broker acts as the enforcement point, granting agents access only for the duration of a specific task, and the product logs agents' actions and communications so teams can see which agent acted on behalf of which user. Palo Alto says agent identity security is also available natively in Prisma AIRS. CyberArk's former 'secure AI agents' URL now redirects to Idira.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "agent-discovery", "agent-privilege-control", "just-in-time-access", "agent-credential-management", "audit-logging" ], "india_relevance": "", "featured": false, "source_url": "https://www.paloaltonetworks.com/idira/agentic", "sources": [ "https://www.paloaltonetworks.com/idira/agentic", "https://www.paloaltonetworks.com/ai-security/agent-identity-security", "https://www.paloaltonetworks.com/about-us/locations", "https://www.paloaltonetworks.com/etc/clientlibs/pan/img/favicons2020/apple-touch-icon-76x76.png" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "workos-authkit-mcp-auth", "name": "WorkOS AuthKit for MCP", "vendor": "WorkOS", "hq_country": "United States", "hq_region": "north-america", "url": "https://workos.com/mcp", "segment": "agent-security", "category": "agent-identity", "categories": [ "agent-identity" ], "summary": "OAuth 2.1 authorization server for MCP servers, so AI agents get scoped, permissioned access to an app without custom auth code.", "description": "WorkOS AuthKit acts as an OAuth 2.1 authorization server for MCP apps following the MCP specification, with PKCE, scopes and tool permissions for fine-grained authorization of agentic workflows. WorkOS Connect can also run as standalone middleware that only provides MCP OAuth flows, without migrating existing user auth.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "mcp-authorization", "agent-authentication", "agent-authorization" ], "india_relevance": "", "featured": false, "source_url": "https://workos.com/mcp", "sources": [ "https://workos.com/mcp", "https://workos.com/legal/terms-of-service", "https://cdn.prod.website-files.com/621f54116cab10f6e9215d8b/621f549024eb2947f6f2f7f7_Favicon%20256x256.png" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "proof-x401-agent-identity", "name": "Proof x401 and agent identity wallet", "vendor": "Proof (Notarize, Inc.)", "hq_country": "United States", "hq_region": "north-america", "url": "https://www.proof.com/use-cases/agentic-web", "segment": "agent-security", "category": "agent-identity", "categories": [ "agent-identity" ], "summary": "Lets a website ask an AI agent to prove whose verified legal identity it acts for and what that person cryptographically approved, via the open x401 protocol and an agent identity wallet.", "description": "Proof launched x401 on Oct 15, 2026: an open HTTP protocol for requesting and verifying evidence of who an AI agent acts for and what the person authorised. The person's agent identity wallet holds their verified legal identity and signs the specific transaction terms on their own device, so the receiving business can check the signature and identity; if the agent changes the amount or recipient, the signature fails. A production MCP server works with assistants that support MCP, and websites can request claims without a sales process, optionally paying per claim over x402.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "agent-authentication", "human-authorization", "verifiable-credentials", "transaction-signing" ], "india_relevance": "", "featured": false, "source_url": "https://www.proof.com/use-cases/agentic-web", "sources": [ "https://www.proof.com/use-cases/agentic-web", "https://www.proof.com/blog/proof-launches-x401-giving-ai-agents-a-way-to-prove-who-they-represent---and-what-theyre-authorized-to-do", "https://x401.proof.com/", "https://www.proof.com/legal/general-terms", "https://cdn.prod.website-files.com/6481e5e765142e9387833b4f/69e15f36c3e1570407de848b_Proof%20Wordmark%20Logo.svg" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "aembit-iam-for-agentic-ai", "name": "Aembit IAM for Agentic AI", "vendor": "Aembit", "hq_country": "United States", "hq_region": "north-america", "url": "https://aembit.io/iam-for-agentic-ai/", "segment": "agent-security", "category": "agent-identity", "categories": [ "agent-identity", "agent-secrets" ], "summary": "Gives AI agents their own identity and policy-governed access to MCP servers and enterprise systems, exchanging credentials on the agent's behalf so agents never hold secrets.", "description": "Aembit implements the MCP OAuth 2.1 authorization code flow so agents and MCP clients, including personal agents, can register, authenticate and receive tokens governed by Aembit Access Policies. It exchanges credentials at request time and issues short-lived, scoped credentials that expire when the task ends, combines the agent's identity with the person it acts for, can revoke an agent instantly, and logs every MCP request with agent identity, user identity, target server and policy decision.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "agent-authentication", "mcp-access-control", "secretless-access", "credential-brokering", "audit-logging" ], "india_relevance": "", "featured": false, "source_url": "https://aembit.io/iam-for-agentic-ai/", "sources": [ "https://aembit.io/iam-for-agentic-ai/", "https://aembit.io/press-release/aembit-raises-25-million-in-series-a-funding-for-non-human-identity-and-access-management/", "https://aembit.io/privacy-policy/", "https://aembit.io/wp-content/uploads/2023/09/aembit-logo-main-no-padding.svg" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "noma-agent-security-governance", "name": "Noma Agent Security and Governance Platform", "vendor": "Noma Security", "hq_country": "United States", "hq_region": "north-america", "url": "https://noma.security/", "segment": "agent-security", "category": "agent-runtime-security", "categories": [ "agent-runtime-security", "agent-governance", "agent-identity" ], "summary": "Discovers and controls AI agents across SaaS, endpoints and cloud, weighing each agent action against context at runtime and able to block, steer, mask data or kill-switch an agent.", "description": "Noma's platform covers SaaS agents, endpoint agents employees use and agents engineers build in the cloud, with AI asset discovery, an agent control plane for policies, agentic identity and access control, MCP security, AI supply-chain security, AI red teaming and AI compliance. At runtime it weighs each agent action against the full session, who the agent acts for, what data it can reach and its normal behaviour, and can stop destructive actions, steer the agent, alert a human, mask data or trigger a kill switch.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "agent-discovery", "agent-runtime-monitoring", "mcp-security", "agent-access-control", "ai-red-teaming", "kill-switch" ], "india_relevance": "", "featured": false, "source_url": "https://noma.security/", "sources": [ "https://noma.security/", "https://noma.security/solutions/mcp-security", "https://noma.security/contact", "https://cdn.prod.website-files.com/6a26af028fd645aaf775e517/6ad63c10e640ccffb4abd167_Website%20Rebrand%20Favicon.png" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "hiddenlayer-agentic-mcp-security", "name": "HiddenLayer Agentic and MCP Security", "vendor": "HiddenLayer", "hq_country": "United States", "hq_region": "north-america", "url": "https://www.hiddenlayer.com/solutions/agentic-mcp-security", "segment": "agent-security", "category": "agent-runtime-security", "categories": [ "agent-runtime-security" ], "summary": "Runtime protection for AI agents and MCP interactions that blocks hidden prompt injection, unsafe memory recall and unsafe tool actions and reconstructs agent activity for investigation.", "description": "HiddenLayer's agentic and MCP security identifies and blocks prompt injection hidden in data, documents, MCP responses or retrieved context before it reaches agent execution, detects unsafe memory recall and cross-agent contamination, and enforces runtime policy across APIs, MCP tools, code execution, communication tools and filesystem operations. It reconstructs agent execution across sessions using LiteLLM proxy interception, SDK instrumentation and gateway inspection, supports OpenAI, AWS Bedrock and MCP-based systems, and can send agent detections to Splunk or Microsoft Sentinel.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "prompt-injection-protection", "agent-runtime-monitoring", "mcp-security", "agent-tracing", "siem-integration" ], "india_relevance": "", "featured": false, "source_url": "https://www.hiddenlayer.com/solutions/agentic-mcp-security", "sources": [ "https://www.hiddenlayer.com/solutions/agentic-mcp-security", "https://www.hiddenlayer.com/news/hiddenlayer-releases-the-2026-ai-threat-landscape-report-spotlighting-the-rise-of-agentic-ai-and-the-expanding-attack-surface-of-autonomous-systems", "https://cdn.prod.website-files.com/694066252c49b2800c70cf57/695e575575d5a8456f341c02_Frame%202147222875%20(1).png" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "microsoft-entra-agent-id", "name": "Microsoft Entra Agent ID", "vendor": "Microsoft", "hq_country": "United States", "hq_region": "north-america", "url": "https://learn.microsoft.com/en-us/entra/agent-id/what-is-microsoft-entra-agent-id", "segment": "agent-security", "category": "agent-identity", "categories": [ "agent-identity", "agent-governance" ], "summary": "Gives AI agents first-class identities in Microsoft Entra so they can be authenticated, authorized, governed and audited like users and workloads, including agents built outside Microsoft.", "description": "Microsoft Entra Agent ID provides agent identities and agent identity blueprints (templates with parent-child relationships) so organisations can apply consistent policies to many agents. It supports OAuth 2.0, MCP and agent-to-agent (A2A) protocols, works with agents built on non-Microsoft platforms such as AWS Bedrock and n8n via the Entra ID Auth SDK or workload identity federation, extends adaptive access, risk detection, lifecycle management and network controls to agents, and logs agent authentication and activity. Microsoft Learn states it is generally available; Microsoft's product page says Agent ID capabilities are available in Microsoft Agent 365 and Microsoft 365 E7 plans.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "agent-authentication", "agent-authorization", "agent-lifecycle-management", "conditional-access", "audit-logging" ], "india_relevance": "", "featured": false, "source_url": "https://learn.microsoft.com/en-us/entra/agent-id/what-is-microsoft-entra-agent-id", "sources": [ "https://learn.microsoft.com/en-us/entra/agent-id/what-is-microsoft-entra-agent-id", "https://learn.microsoft.com/en-us/entra/agent-id/whats-new-agent-id", "https://www.microsoft.com/en-us/security/business/identity-access/microsoft-entra-agent-id", "https://news.microsoft.com/facts-about-microsoft/", "https://uhf.microsoft.com/images/microsoft/RE1Mu3b.png" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "auth0-for-ai-agents", "name": "Auth0 for AI Agents (AI identity, Token Vault)", "vendor": "Okta (Auth0)", "hq_country": "United States", "hq_region": "north-america", "url": "https://auth0.com/ai/docs/overview", "segment": "agent-security", "category": "agent-identity", "categories": [ "agent-identity", "agent-secrets" ], "summary": "Developer identity layer for AI agents: user login, on-behalf-of API access, a Token Vault that holds third-party tokens for agents, and asynchronous human approval for sensitive actions.", "description": "Auth0's AI docs describe adding user authentication to interactive and headless AI agents with OAuth 2.0 and OpenID Connect, letting agents call first-party APIs on the user's behalf without broad access, and connecting agents to services such as Google, Slack or GitHub through Token Vault, which obtains, stores and refreshes API tokens so agents never handle the keys. Using CIBA, agents can request user approval for sensitive operations by push, SMS or email.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "agent-authentication", "delegated-authorization", "token-vault", "human-in-the-loop-approval" ], "india_relevance": "", "featured": false, "source_url": "https://auth0.com/ai/docs/overview", "sources": [ "https://auth0.com/ai/docs/overview", "https://auth0.com/solutions/ship-ai-agents", "https://www.okta.com/contact/", "https://www.okta.com/company/", "https://cdn.auth0.com/website/website/favicons/auth0-favicon.svg" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "aggate-ai-agent-governance", "name": "Aggate AI agent governance platform", "vendor": "Aggate (Aggate SAS)", "hq_country": "France", "hq_region": "europe", "url": "https://aggate.ai/", "segment": "agent-security", "category": "agent-governance", "categories": [ "agent-governance", "agent-runtime-security", "agent-identity" ], "summary": "Checks, authorises and traces every action an AI agent takes before it reaches company systems, weighing the agent's identity, intent and context, and keeps an audit trail.", "description": "Aggate (French-language site) evaluates each agent action at execution time on three questions: which agent is acting and for whom (identity), whether the action matches what it was asked to do (intent), and whether the environment makes it risky (context, e.g. test vs production). Actions are allowed, sent for human validation or blocked, and the trace is kept for auditors; teams can connect an agent to see what it could reach and what Aggate would decide. The site maps the product to EU AI Act, DORA, NIS 2, GDPR and ISO/IEC 42001 obligations and says it is available on a marketplace.", "pricing": "", "pricing_model": "", "frameworks": [ "dora" ], "frameworks_claimed": [ "EU AI Act (Regulation (EU) 2024/1689)", "DORA", "NIS 2", "GDPR", "ISO/IEC 42001" ], "tasks": [ "agent-authorization", "intent-verification", "human-in-the-loop-approval", "audit-logging" ], "india_relevance": "", "featured": false, "source_url": "https://aggate.ai/", "sources": [ "https://aggate.ai/", "https://aggate.ai/mentions-legales", "https://aggate.ai/favicon.svg" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "unosecur-ai-agent-security", "name": "Unosecur AI Agent Security and Freeze Agent", "vendor": "Unosecur (Unosecur GmbH)", "hq_country": "Germany", "hq_region": "europe", "url": "https://www.unosecur.com/agentic-ai-security", "segment": "agent-security", "category": "agent-identity", "categories": [ "agent-identity", "agent-runtime-security" ], "summary": "Discovers AI agents across cloud and SaaS, governs them as identities alongside humans and machine identities, traces their access paths and offers a kill switch for rogue agents.", "description": "Unosecur brings AI agents into its unified identity fabric with human and non-human identities: it continuously identifies agents across the cloud and SaaS stack (integrations include Amazon Bedrock, Microsoft Copilot Studio, Google Vertex AI, Microsoft 365 Copilot and Salesforce Agentforce), traces each agent through its tools to the service roles and data it can reach, pinpoints the access an agent actually needs and prioritises risk by asset criticality. Unosecur Freeze Agent is a kill switch that stops rogue or compromised AI agents in seconds.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "agent-discovery", "agent-access-governance", "least-privilege-enforcement", "kill-switch" ], "india_relevance": "", "featured": false, "source_url": "https://www.unosecur.com/agentic-ai-security", "sources": [ "https://www.unosecur.com/agentic-ai-security", "https://www.unosecur.com/agentic-ai-mcp", "https://www.unosecur.com/privacy-policy", "https://cdn.prod.website-files.com/6971b283ecf17527dee7dd85/6971b66986b690a5111da70f_Unosecur%20Logo%20(%20For%20Light%20Background)%201.svg" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "identos-policyarc", "name": "IDENTOS PolicyArc", "vendor": "IDENTOS", "hq_country": "Canada", "hq_region": "north-america", "url": "https://www.identos.com/news/identos-launches-policyarc-ai-ready-access-control", "segment": "agent-security", "category": "agent-identity", "categories": [ "agent-identity", "agent-governance" ], "summary": "Access-control service that lets people delegate a scoped, time-limited slice of their own access to an AI agent, revoked automatically when the task ends, with every decision on record.", "description": "IDENTOS announced PolicyArc as generally available: a single place to manage, enforce and audit access policies across applications and extend them safely to AI agents acting on people's behalf. Users can grant an agent a scoped, time-limited version of their own access that is revoked automatically when the task or window ends, and every access decision is recorded for compliance teams.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "delegated-authorization", "agent-authorization", "time-bound-access", "audit-logging" ], "india_relevance": "", "featured": false, "source_url": "https://www.identos.com/news/identos-launches-policyarc-ai-ready-access-control", "sources": [ "https://www.identos.com/news/identos-launches-policyarc-ai-ready-access-control", "https://policyarc.com/", "https://www.identos.com/contact-us", "https://cdn.prod.website-files.com/699d535776f44f2f8aa4309f/6aa35cd59a554a649d69408e_identos-logo.svg" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "kosmoy-ai-agent-management", "name": "Kosmoy AI Agent Management", "vendor": "Kosmoy (Libeccio Ventures S.r.l.)", "hq_country": "Italy", "hq_region": "europe", "url": "https://www.kosmoy.com/platform/ai-agent-management/", "segment": "agent-security", "category": "agent-governance", "categories": [ "agent-governance", "agent-runtime-security" ], "summary": "Registers, observes, governs and contains AI agents across Microsoft, AWS, Google, Salesforce, ServiceNow and private runtimes, with a policy gateway, short-lived credentials and a kill switch.", "description": "Kosmoy keeps every agent in a master registry matched against approved use cases, with an identity and named owner per agent and shadow AI flagged. Agent traffic is routed through one policy point (RBAC, guardrails, budgets) over A2A alongside LLM and MCP calls; Mission Control issues short-lived credentials and can stop any agent mid-run, and agents run inside an Action Capsule sandbox get full observability, guardrails and a kill switch. For agents on third-party platforms it provides inventory, risk classification and observability where telemetry allows.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "agent-inventory", "agent-governance", "ai-gateway", "kill-switch", "audit-logging" ], "india_relevance": "", "featured": false, "source_url": "https://www.kosmoy.com/platform/ai-agent-management/", "sources": [ "https://www.kosmoy.com/platform/ai-agent-management/", "https://www.kosmoy.com/platform/ai-gateway/", "https://www.kosmoy.com/privacy-policy/", "https://www.kosmoy.com/images/kosmoy-logo-light.svg" ], "last_checked": "2026-10-29", "added": "2026-10-29" }, { "slug": "shufti-know-your-agent", "name": "Shufti Know Your Agent (KYA)", "vendor": "Shufti (Shufti Pro Limited)", "hq_country": "United Kingdom", "hq_region": "europe", "url": "https://shuftipro.com/ai-agent-verification/", "segment": "agent-security", "category": "agent-identity", "categories": [ "agent-identity", "agent-governance" ], "summary": "AI agent verification for businesses that receive requests from AI agents: it checks which agent is acting, confirms that a real, verified person authorised it, and enforces that person's mandate before the agent can access data or trigger a transaction.", "description": "Per Shufti's Oct 28, 2026 launch release and its AI agent verification page, KYA combines agent recognition, trust scoring, spoof-resistant identity verification and permission checks in one flow. A principal sets a mandate for each agent (permitted actions, limits, expiry, revocation) that is enforced as a policy on every request; requests outside the mandate or above a risk threshold trigger human-in-the-loop control with a selfie liveness check by the authorising person, and agent actions stay traceable. Shufti positions it for agentic commerce and other AI agent workflows (e.g. blocking agents from placing bets in iGaming) and says KYA is now available.", "pricing": "", "pricing_model": "", "frameworks": [], "frameworks_claimed": [], "tasks": [ "agent-authentication", "agent-authorization", "human-authorization", "audit-logging" ], "india_relevance": "", "featured": false, "source_url": "https://shuftipro.com/press-release/shufti-launches-know-your-agent-to-verify-ai-agents/", "sources": [ "https://shuftipro.com/press-release/shufti-launches-know-your-agent-to-verify-ai-agents/", "https://shuftipro.com/ai-agent-verification/", "https://shuftipro.com/contact-us/" ], "last_checked": "2026-10-30", "added": "2026-10-30" } ] }