Aurascape is a Guardrails / LLM Firewall vendor based in Santa Clara, California, USA. Its flagship product is Aurascape AI security platform (AI Proxy, Zero-Bypass MCP Gateway, Coding Assistant Guardrails, AI Discovery & Risk Scoring). Against the OWASP LLM Top 10 (2026) we score it Covered on 3 of 10 risks, Partial on 2, with 6 MITRE ATLAS techniques mapped as of . It's independent.
- 3/10
- Risks covered
- 2
- Partial
- 6
- ATLAS techniques
- $50M
- Funding
- Active
- Status
At a glance
- Category
- Guardrails / LLM Firewall
- Headquarters
- Santa Clara, California, USA
- Founded
- 2024
- Employees
- 51-100
- Flagship product
- Aurascape AI security platform (AI Proxy, Zero-Bypass MCP Gateway, Coding Assistant Guardrails, AI Discovery & Risk Scoring)
- Deployment
- Inline proxy and endpoint clients; Zero-Bypass MCP Gateway deploys by prepending one prefix to existing MCP server URLs
- Best for
- Enterprises that need inline visibility and data protection over employee AI use, copilots and coding assistants, with MCP governance added
- Funding to date
- $50M
- Last round
- Undisclosed · 2025-04
- Status
- Active
Facts as of .
What does Aurascape do?
Sits inline on employee and agent AI traffic: classifies every prompt and response, and governs tool calls through an MCP gateway. It ships as Inline proxy and endpoint clients; Zero-Bypass MCP Gateway deploys by prepending one prefix to existing MCP server URLs, and it's built for enterprises that need inline visibility and data protection over employee AI use, copilots and coding assistants, with MCP governance added. It's a weaker fit for teams needing vector-store controls, training-data controls or token spend limits; none is documented.
Which OWASP LLM Top 10 risks does Aurascape cover?
Aurascape's strongest verdict is LLM01 Prompt Injection. The two tables below come from our editorial read of Aurascape's public documentation as of (it's what the docs support, not a hands-on lab test), and this row also sits in the full matrix across all 27 vendors.
| OWASP LLM Risk | What it is | Aurascape coverage | How it's addressed | Source |
|---|---|---|---|---|
| LLM01 Prompt Injection | User or hidden input overrides the model's rules or intended behavior. | Covered | Runtime guardrails in the AI Proxy detect prompt injection, and prompts and instructions are policy-enforced in real time. | https://aurascape.ai/secure-agentic-ai/ |
| LLM02 Sensitive Information Disclosure | The model leaks secrets, personal data, or proprietary content in its output. | Covered | The inline classification engine discovers and classifies sensitive data in real time across AI interactions, in allow and block modes. | https://aurascape.ai/safeguard-ai-use/ |
| LLM03 Excessive Agency | An agent holds more permissions, tools, or autonomy than the task needs. | Covered | A custom registry of approved MCP servers and role-based endpoints blocks unsanctioned servers automatically, for gateway-routed traffic. | https://aurascape.ai/secure-agentic-ai/ |
| LLM04 Supply Chain | Compromised models, datasets, plugins, or dependencies add risk before runtime. | Partial | Coding Assistant Guardrails stop typosquatted and poisoned packages, but no model-file or dataset provenance control is documented. | https://aurascape.ai/coding-assistant-guardrails/ |
| LLM05 Data and Model Poisoning | Tampered training or fine-tuning data corrupts how the model behaves. | Unverified | No control over training or fine-tuning data found on Aurascape product pages as of 2026-10-30. | None found |
| LLM06 Unbounded Consumption | Uncontrolled requests drive cost, denial of service, or model extraction. | Unverified | No rate limit, token budget or cost control found on Aurascape product pages as of 2026-10-30. | None found |
| LLM07 Misinformation | The model produces false or fabricated content that users act on. | Unverified | No hallucination or factual-accuracy control found on Aurascape product pages as of 2026-10-30. | None found |
| LLM08 Hidden Context Exposure | Hidden instructions, retrieved policy text, or tool schemas get extracted, inferred, or reconstructed from the model. | Unverified | No named system-prompt or hidden-context protection found on Aurascape product pages as of 2026-10-30. | None found |
| LLM09 Vector and Embedding Weaknesses | Flaws in RAG stores let attackers poison, extract, or infer data. | Unverified | No vector-store or embedding control found on Aurascape product pages as of 2026-10-30. | None found |
| LLM10 Improper Output Handling | Downstream systems trust model output without checking it, enabling injection or code execution. | Partial | Runtime guardrails list output sanitization and code-injection checks, but downstream handling detail is not published. | https://aurascape.ai/secure-agentic-ai/ |
| ATLAS Technique (ID) | Tactic | Aurascape coverage | Notes | Source |
|---|---|---|---|---|
| LLM Prompt Injection (AML.T0051) | Execution | Covered | Runtime guardrails detect prompt injection, with policy enforced inline through the AI Proxy. | https://aurascape.ai/secure-agentic-ai/ |
| LLM Jailbreak (AML.T0054) | Defense Evasion / Privilege Escalation | Covered | Runtime guardrails detect jailbreaks inline, and pre-deployment testing simulates jailbreak attempts. | https://aurascape.ai/secure-agentic-ai/ |
| Exfiltration via AI User Interface (AML.T0136) | Exfiltration | Covered | Inline data classification and fingerprinting inspect prompts and responses and can block in real time. | https://aurascape.ai/safeguard-ai-use/ |
| AI Agent Tool Invocation (AML.T0053) | Execution / Privilege Escalation / Lateral Movement | Covered | The MCP Gateway verifies each gateway-routed tool call before it reaches an external system. | https://aurascape.ai/secure-agentic-ai/ |
| AI Agent Tool Poisoning (AML.T0110) | Persistence | Covered | Aurascape inspects tool metadata exchanges and blocks the poisoned tool before data leaves. | https://aurascape.ai/secure-agentic-ai/ |
| Supply Chain Compromise (AML.T0010) | Initial Access | Partial | Coding Assistant Guardrails stop poisoned and typosquatted packages, but model and dataset supply chain controls are not documented. | https://aurascape.ai/coding-assistant-guardrails/ |
Which OWASP Agentic risks does Aurascape cover?
3 of 10 Covered and 2 Partial against the OWASP Top 10 for Agentic Applications, scored from public documentation as of 2026-10-30.
| Agentic Risk | What it is | Aurascape coverage | How it's addressed | Source |
|---|---|---|---|---|
| Agent Goal Hijack (ASI01) | Attackers redirect what an agent is trying to do, often through documents or other content it reads. | Covered | Runtime guardrails detect prompt injection in the model conversation, and the MCP Gateway blocks the tool actions that follow. | https://aurascape.ai/secure-agentic-ai/ |
| Tool Misuse and Exploitation (ASI02) | An agent uses a legitimate tool in a harmful way, or a manipulated tool call does damage. | Covered | The MCP Gateway verifies gateway-routed tool calls against approved tools and blocks unapproved ones by default. | https://aurascape.ai/secure-agentic-ai/ |
| Identity and Privilege Abuse (ASI03) | Agents inherit, reuse, or escalate credentials and permissions beyond what the task needs. | Partial | Aurascape enforces boundaries on agent identity and tool access for built agents, but delegated credential controls are not detailed. | https://aurascape.ai/secure-agentic-ai/ |
| Agentic Supply Chain Vulnerabilities (ASI04) | Compromised MCP servers, plugins, skills, or agents enter the runtime. | Partial | Aurascape catalogs MCP servers with a risk score and security scan and blocks poisoned tools, but skills and plugins are not documented. | https://aurascape.ai/secure-agentic-ai/ |
| Unexpected Code Execution (ASI05) | An agent generates or runs code that gives an attacker execution. | Covered | Coding Assistant Guardrails stop destructive or exploitative commands before they reach the shell. | https://aurascape.ai/coding-assistant-guardrails/ |
| Memory and Context Poisoning (ASI06) | Attackers plant content in an agent's memory or context that shapes later decisions. | Unverified | No control for agent memory or context poisoning found on Aurascape product pages as of 2026-10-30. | None found |
| Insecure Inter-Agent Communication (ASI07) | Messages between agents get spoofed, intercepted, or tampered with. | Unverified | No inter-agent message authentication or integrity control found on Aurascape product pages as of 2026-10-30. | None found |
| Cascading Failures (ASI08) | One fault spreads across connected agents and systems. | Unverified | Aurascape traces data across chained tool calls, but no containment of failures spreading across agents is documented. | https://aurascape.ai/secure-agentic-ai/ |
| Human-Agent Trust Exploitation (ASI09) | An agent's confident output talks a person into approving something harmful. | Unverified | Policy can hold a risky command for confirmation, but no control against manipulated human approval is named. | https://aurascape.ai/coding-assistant-guardrails/ |
| Rogue Agents (ASI10) | An agent drifts or is compromised and acts outside policy. | Unverified | Aurascape flags tools operating outside the gateway, but no rogue-agent behavior detection is documented. | https://aurascape.ai/secure-agentic-ai/ |
Is Aurascape independent, and how is it funded?
Aurascape is an independent company as of . It has raised $50M to date dated 2025-04. Lead investors: Mayfield Fund, Menlo Ventures.
Aurascape alternatives
The closest alternatives we track in Guardrails / LLM Firewall are Enkrypt AI, Lakera, Lasso Security. On the open-source side, Guardrails covers similar ground.
Frequently asked questions
What is Aurascape used for?
Aurascape is a Guardrails / LLM Firewall vendor. Its flagship product is Aurascape AI security platform (AI Proxy, Zero-Bypass MCP Gateway, Coding Assistant Guardrails, AI Discovery & Risk Scoring). It ships as Inline proxy and endpoint clients; Zero-Bypass MCP Gateway deploys by prepending one prefix to existing MCP server URLs, and it's built for enterprises that need inline visibility and data protection over employee AI use, copilots and coding assistants, with MCP governance added.
Is Aurascape independent or acquired?
Aurascape is an independent company as of 2026-10-30 and has not been acquired.
How many OWASP LLM Top 10 risks does Aurascape cover?
We score Aurascape as Covered on 3 of the ten OWASP LLM Top 10 (2026) risks, with 2 more Partial, as of 2026-10-30. The table on this page breaks down every risk, including the ones marked Not covered or Unverified.
How is Aurascape deployed?
Aurascape ships as Inline proxy and endpoint clients; Zero-Bypass MCP Gateway deploys by prepending one prefix to existing MCP server URLs.
What are the best alternatives to Aurascape?
The closest Guardrails / LLM Firewall alternatives llmthreat tracks are Enkrypt AI, Lakera, Lasso Security. On the open-source side, Guardrails covers similar ground.