Aurascape

No entries match these filters. Clear filters

1. September New Innovations Recap: More Precise Protection for AI Work (2026-10-05) Oct 5, 2026 · issue 097

   Aurascape September 2026 adds prompt masking in Protection policy, classification management APIs, SSOSSOSingle sign-on, using one identity provider to sign in to multiple applications. It reduces separate passwords and centralizes account and access management. audit logging, and richer on-demand reporting.

   • Adds sensitive data masking as a mode of the redact action in Protection policy, replacing sensitive values in prompts before they reach the LLMLLMA large language model, a neural network trained on large amounts of text to recognize patterns and generate language. Applications use it for tasks such as answering, summarizing and writing code. and restoring real values in the response the user reads.
   • New APIs automate management of data patterns, data categories, and data combination lists for data classification configurations, and can retrieve the supporting objects those configurations reference.
   • Adds administrator SSO activity to Audit Logs, recording successful and failed login attempts with timestamp, status, and source IP address.
   • On-demand reports now include a Unique User Counts by Application metric and a Top Protection Policy Hits bar chart alongside the existing application breakdown.
   • Conversation records for coding sessions now separate system prompt, user prompt, and endpoint details into individual fields.

   +2 more show less

   • Adds more options for administrators to customize the frequency of real-time coaching notifications when users repeatedly trigger the same Protection policy, to reduce alert fatigue.
   • Administrators can unmask sensitive values in the Conversations view (subject to user role permissions) during investigations.

   7 details in Everything

   • Updated integration Granular model-level access policy for Claude, Anthropic API, ChatGPT, OpenAI API, Google Vertex AI API, and Amazon Bedrock
   • Updated integration Expanded MCP conversation record detail via Zero-Bypass MCP Gateway
2. Secure the Coding Session, Not Just the Code (2026-09-04) Sep 4, 2026 · issue 065

   Aurascape launches Coding Assistant Guardrails to govern live AI coding sessions — blocking secrets, destructive commands, and prompt injectionprompt injectionAn attack where text the model reads, such as a web page, file or tool result, carries instructions that override the user's. It is the main security risk for agents with tools. in flight.

   • Introduces Coding Assistant Guardrails, a new solution that inspects and enforces policy on live AI coding-assistant sessions — covering Claude Code, GitHub Copilot, and Cursor — before any artifact exists to scan.
   • Decodes coding-assistant streaming protocols (Server-Sent Events, WebSockets, Protobuf, gRPCgRPCAn open-source remote procedure call framework from Google that uses HTTP/2 and Protocol Buffers to let services call each other's functions across a network, giving tools low-latency, strongly typed inter-process communication., and MCPMCPModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface.) inline without buffering, preserving native streaming speed for developers while enabling content inspection.
   • Detects and blocks or redacts secrets and credentials pasted into prompts in flight, before the model receives them, with the event captured as auditable evidence.
   • Intercepts destructive shell commands and unauthorized git pushes at the wire level, in the tool call the model streams back, before the client executes them — with configurable block or confirmation-checkpoint policy.
   • Detects prompt injection hidden in READMEs, documentation, and MCP outputs, neutralizing hijacked instructions before the agent acts on them.

   +6 more show less

   • Flags typosquatted packages and spoofed links recommended by a model, surfacing supply-chain attempts as visible, reportable events.
   • Provides continuous discovery of IDE plugins, extensions, and MCP servers as they appear, with new or uncataloged apps receiving signatures on a 48-hour SLASLAA formal agreement between a service provider and customer that defines expected uptime, response times, and remedies for failures, giving cyber tools a contractual baseline for availability and support commitments. — removing the need for hand-maintained allowlists.
   • Exposes MCP tool calls down to their parameters for visibility and policy enforcement.
   • Delivers auditable conversation logs — capturing prompts, responses, connections, and tool calls behind each policy flag — under role-based access controlrole-based access controlRole-based access control, where permissions are assigned to roles such as reader or administrator and users receive those roles. It makes access easier to manage than individual permission lists. for security investigations.
   • Offers customizable coaching templates that surface in-browser, in-IDE, or in-terminal explanations and safer paths when a policy action fires.
   • Enables natural-language queries over AI usage, risk, and policy data through Auri.

   11 details in Everything

3. Secure the Coding Session, Not Just the Code (2026-09-04) Sep 4, 2026 · issue 011

   Aurascape launches Coding Assistant Guardrails to govern live AI coding sessions — blocking secrets, destructive commands, and prompt injectionprompt injectionAn attack where text the model reads, such as a web page, file or tool result, carries instructions that override the user's. It is the main security risk for agents with tools. in flight.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

   • Introduces Coding Assistant Guardrails, a new solution that inspects and enforces policy on live AI coding-assistant sessions — covering Claude Code, GitHub Copilot, and Cursor — before any artifact exists to scan.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
   • Decodes coding-assistant streaming protocols (Server-Sent Events, WebSockets, Protobuf, gRPCgRPCAn open-source remote procedure call framework from Google that uses HTTP/2 and Protocol Buffers to let services call each other's functions across a network, giving tools low-latency, strongly typed inter-process communication., and MCPMCPModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface.) inline without buffering, preserving native streaming speed for developers while enabling content inspection.
   • Detects and blocks or redacts secrets and credentials pasted into prompts in flight, before the model receives them, with the event captured as auditable evidence.
   • Intercepts destructive shell commands and unauthorized git pushes at the wire level, in the tool call the model streams back, before the client executes them — with configurable block or confirmation-checkpoint policy.
   • Detects prompt injection hidden in READMEs, documentation, and MCP outputs, neutralizing hijacked instructions before the agent acts on them.

   +6 more show less

   • Flags typosquatted packages and spoofed links recommended by a model, surfacing supply-chain attempts as visible, reportable events.
   • Provides continuous discovery of IDE plugins, extensions, and MCP servers as they appear, with new or uncataloged apps receiving signatures on a 48-hour SLASLAA formal agreement between a service provider and customer that defines expected uptime, response times, and remedies for failures, giving cyber tools a contractual baseline for availability and support commitments. — removing the need for hand-maintained allowlists.
   • Exposes MCP tool calls down to their parameters for visibility and policy enforcement.
   • Delivers auditable conversation logs — capturing prompts, responses, connections, and tool calls behind each policy flag — under role-based access controlrole-based access controlRole-based access control, where permissions are assigned to roles such as reader or administrator and users receive those roles. It makes access easier to manage than individual permission lists. for security investigations.
   • Offers customizable coaching templates that surface in-browser, in-IDE, or in-terminal explanations and safer paths when a policy action fires.
   • Enables natural-language queries over AI usage, risk, and policy data through Auri.

   11 details in Everything

4. Zero-Bypass MCP Gateway Data Sheet (2026-08-20) Aug 20, 2026 · issue 054

   Aurascape Zero-Bypass MCP GatewayMCP GatewayA server that sits between MCP clients and one or more Model Context Protocol backends, routing tool calls and aggregating context sources so a single endpoint exposes many AI-agent integrations at once. pairs an MCP Gateway with an AI Proxy to block tool calls that skip policy enforcement

   • Introduces the Zero-Bypass MCPMCPModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface. Gateway, which governs MCP tool access and marks every approved call so the AI Proxy can block any tool invocation that bypassed the Gateway before the model acts.
   • Adds automatic MCP server and tool discovery with automatic enrollment, eliminating manual registration of AI agent tool surfaces.
   • Supports tool-level sanctioning bound to users and groups, enabling fine-grained access control over which agents can call which MCP tools.
   • Provides policy actions ranging from coaching to hold-for-approval to blocking for MCP tool calls.
   • Implements OAuth 2.1OAuth 2.1A consolidated draft of the OAuth authorization framework that merges OAuth 2.0 with its most important security BCP updates, dropping unsafe flows so tools can delegate user access without handling credentials directly. authorization with tokens scoped per user per server for MCP connections.

   +2 more show less

   • Generates audit-ready records of every observed MCP conversation for compliance and investigation.
   • Detects and maps coverage against the OWASPOWASPA nonprofit foundation that publishes open standards, guides, and ranked vulnerability lists—most famously the OWASP Top 10—giving cyber tools a vendor-neutral benchmark to test or report against. MCP Top 10, including tool-poisoning attacks and credential-chaining scenarios across multiple systems.

   7 details in Everything