# Changelog

### April 2026

#### Application Enhancements

Unified MCP and LLM Gateway into a single AI Gateway experience with a shared playground and policy model. Added AWS Bedrock support, expanded LLM provider management, and improved guardrail visibility for blocked or redacted responses. Enhanced Ramparts scanning with SARIF output, OWASP MCP taxonomy mapping, dependency scanning, replay support, and repository-wide config discovery. Improved Observatory session visibility and overall scanning workflows.

#### Security & Compliance

Fixed multiple tenant isolation and privilege escalation vulnerabilities affecting API key access and revocation flows. Strengthened JWT validation, unified Cedar policy namespaces, and improved Shield policy visibility in user-facing responses. Resolved configuration merge and secrets preservation issues, and removed accidental binary artifacts from source control.

#### Performance Improvements

Removed unused modules and dependencies to reduce binary size, compile times, and attack surface. Optimized dependency vulnerability scanning with parallel execution to improve scan efficiency without added latency.

#### DevOps & Infrastructure

Added a comprehensive AI Gateway E2E test suite covering security and streaming scenarios. Standardized proxy URL formats across services, improved OAuth compatibility, automated release changelog generation, and bundled CA certificates to improve TLS reliability across environments.

### March 2026

#### Application Enhancements

Updated **Observatory** — a unified security intelligence console in Highflame Studio that aggregates signals from every Highflame product into a single interface. Observatory includes six views: Command Center (security posture score, UEBA entity risk ranking, blast radius map, coverage mesh), Threats (cross-product event table with faceted search and policy gap detection), Sessions (end-to-end agent session timelines with gateway and SDK event correlation), Traces (OpenTelemetry distributed trace waterfall visualization), Tools & Agents (tool usage analytics and MCP server inventory), and Playground (interactive policy testing with a predefined attack scenario library and live event streaming).

Updated **Browser Security** — enterprise browser extension that detects and enforces policy against AI-targeted threats at the browser level. Supports Chrome, Edge, Firefox, and Safari via managed enterprise deployment (Chrome Browser Cloud Management, Intune, Group Policy, MDM). Threat coverage includes prompt injection, data exfiltration, token theft, sensitive file uploads, clipboard attacks, XSS/script injection, and storage protection. Targeted detections run locally in the browser kernel (12.7 KB). Violations surface in Highflame Studio with real-time alerts, device inventory, and threat analytics. Browser violations are correlated into Observatory Sessions for cross-product investigation.

Added **TypeScript framework adapters** to the Shield SDK — LangGraph, CrewAI, AWS Strands, and Azure AI Foundry integrations are now available in `@highflame/sdk/integrations/*`, matching the existing Python adapter surface. All framework integrations are namespaced under `@highflame/sdk/integrations/<name>` (TypeScript) and `highflame.integrations.<name>` (Python).

#### Security & Compliance

Introduced **structured OAuth2 error types** across the ZeroID server and Python/TypeScript SDKs. All token endpoint errors now map to typed error classes (`InvalidClientError`, `InvalidGrantError`, `InsufficientScopeError`, etc.) for precise error handling in agent authorization flows. Added `/oauth2/token/verify` forward-auth endpoint for auth proxy integration.

Added **local JWT verification** to the `authjwt` client package — agents can now validate tokens in-process against a JWKS endpoint without a round-trip to ZeroID, reducing authorization latency and external dependency on the hot path.

#### Performance Improvements

Improved token verification path with in-process JWKS caching and lazy key rotation, eliminating redundant network calls on each token check. Optimized Shield SDK span emission to batch low-priority telemetry events and avoid blocking LLM call latency.

#### DevOps & Infrastructure

Added integration test suite for ZeroID auth code grant flow, covering token issuance, refresh, and revocation end-to-end. Resolved Trivy scan failures in CI across SDK and server containers. Updated framework integration peer dependency minimums to align with stable releases: LangGraph `>=1.0`, CrewAI `>=1.0`, Strands Agents `>=1.0`, `@langchain/langgraph >=1.0.0`, `@aws/bedrock-ai-agents-strands >=1.0.0`.

### February 2026

#### Application Enhancements

Improved event visibility and threat visualization with better filtering and severity indicators. Added an MCP Gateway inspector for interactive testing and enhanced multi-tenant policy handling with caching. Improved Red Team usability and scan filtering.

#### Security & Compliance

Strengthened multi-tenancy isolation and credential handling, ensuring no sensitive data is persisted. Improved protection against injection and XSS risks, and enhanced security metrics tracking.

#### Performance Improvements

Reduced code complexity, improved concurrent policy access, and added fallback mechanisms for reliability. Fixed timeout issues and optimized UI performance.

#### DevOps & Infrastructure

Enhanced CI/CD and deployment workflows, improved health monitoring, and updated container and model configurations.

### January 2026

#### Application Enhancements

Launched Overwatch UI with monitoring, session inspection, and policy views. Enhanced policy management and added agent activity insights. Expanded MCP integration and introduced standardized error handling with improved SDK resilience and documentation.

#### Security & Compliance

Fixed multiple critical vulnerabilities and implemented multi-tenancy with data isolation. Introduced advanced content attack detection (Doublespeak Engine) and strengthened audit logging, validation, and credential protection.

#### Performance Improvements

Improved logging, database efficiency, and connection management. Added caching and enhanced health checks for better system reliability.

#### DevOps & Infrastructure

Enhanced CI/CD workflows, improved container builds, expanded test coverage, and strengthened production stability and migration processes.

### December 2025

#### Application Enhancements

Introduced Model Scan as a global feature and launched a new Red Team Dashboard with risk and regression tracking. Added system-level analytics and threat alerts, improved gateway navigation and UI, and enabled tool calling across supported language providers.

#### Security & Compliance

Added policy compliance mapping, enhanced YARA-based threat scanning, improved MCP gateway security, and completed Palisade launch readiness.

#### Performance Improvements

Optimized attack detection engines, improved API polling and caching, and streamlined gateway and workspace selection for faster response times.

#### DevOps & Infrastructure

Stabilized CI pipelines and smoke tests, improved gateway health monitoring, enabled automatic gateway selection, and introduced reusable dashboard components for consistency.

### November 2025

#### Application Enhancements

Added system-level data protection and threat alert views with unified monitoring across gateways. Improved navigation and policy selection flows. Enhanced gateway and MCP support, including sub-registry functionality and better model handling.

#### Security & Compliance

Introduced feature-flagged access to new security capabilities. Strengthened threat detection and DLP processing, added YARA-based scanning, and improved ASCII language checks. Enhanced secrets management and security query handling.

#### Performance Improvements

Improved Redis and buffer handling to support high-load scenarios. Optimized database queries and trace processing for lower latency and improved reliability.

#### DevOps & Infrastructure

Stabilized E2E and smoke testing for security and policy workflows. Improved CI quality checks, and streamlined integration deployments. Completed the platform rebrand to Highflame and improved CI workflows with stronger linting and quality checks.

### October 2025

#### Application Enhancements

Added user invitations with role-based access and improved Code Agent policy management. Expanded guardrail processor support and enhanced Chronicle views with a reusable architecture for better consistency and navigation.

#### Security & Compliance

Improved processor safety and validation to prevent race conditions and runtime errors. Strengthened secrets management and upgraded to Go 1.25.3 with critical security fixes. Enhanced YARA-based threat detection.

#### Performance Improvements

Optimized concurrency and locking behavior to improve throughput. Improved frontend rendering and form handling for better responsiveness and maintainability.

#### DevOps & Infrastructure

Enhanced container management in the MCP Gateway. Improved smoke tests and CI quality checks, updated ModelSpec versions, and fixed Content Security Policy (CSP) issues.

### September 2025

#### Application Enhancements

Added Business Unit–level policy management for centralized governance. Enhanced Red Team reporting with richer metrics and visualizations. Introduced a comprehensive MCP security proxy and engine metrics APIs for better visibility into model performance.

#### Security & Compliance

Fixed critical security bypasses and strengthened DLP template validation. Improved secrets management and enabled feature-flagged security rollouts. Enhanced dangerous tool and injection detection across security proxies and guardrails.

#### Performance Improvements

Improved DLP processor efficiency, enhanced Redis compatibility, and enabled Go 1.25 runtime optimizations. Increased MCP security proxy throughput and latency performance and added caching to reduce repeated validation costs.

#### DevOps & Infrastructure

Expanded E2E and unit test coverage for Red Team and application auditing. Optimized Docker images, improved cloud Redis and database connectivity, strengthened code quality, and enhanced technical documentation.

### August 2025

#### Application Enhancements

Added internationalization (i18n) support and improved Red Team assessments and reporting. Expanded MCP capabilities with policy, tools, and security management, along with clearer application and tool visualizations. Improved UI performance and navigation across traces and chronicles.

#### Security & Compliance

Strengthened security analysis and governance with consolidated results, enhanced auditing, and richer diagnostics. Added advanced threat detection for tools and prompts, with clearer visibility into security issues and severity.

#### Performance Improvements

Improved security scanning performance through parallel processing and caching. Enhanced runtime stability and logging, fixed concurrency issues, and improved session and resource management.

#### DevOps & Infrastructure

Streamlined CI/CD pipelines and deployments, improved Kubernetes and Azure integration, enhanced E2E test coverage, and optimized pipeline execution and scheduling.

### July 2025

#### Application Enhancements

Expanded MCP tooling with dynamic management and improved Red Team capabilities. Added advanced content analysis including hallucination, sentiment, and fact-checking. Improved analytics filtering, pagination, and support for non-English content detection.

#### Security & Compliance

Improved authentication, CORS handling, and added guardrail error metrics with auto-provisioned security policies.

#### Performance Improvements

Improved request reliability with timeouts, async processing, retries, and better streaming handling. Optimized analytics queries, execution chains, and added storage purging policies with enhanced monitoring.

#### DevOps & Infrastructure

Upgraded to Go 1.24.4 and strengthened CI/CD with Trivy scans. Improved cloud deployments, Redis authentication, secret management, and faster E2E testing.

### June 2025

#### Application Enhancements

Added feature flags for Red Team capabilities and improved content detection with fuzzy matching and format preservation. Enhanced multi-language support and improved model trace metadata visibility. Strengthened ASCII-based language detection for better accuracy.

#### Security & Compliance

Improved DLP and security guardrails, including advanced pattern matching and detection of hidden or obfuscated content. Removed legacy X-API-Key header support to strengthen security. Added format-preserving anonymization for sensitive data.

#### Performance Improvements

Improved database reliability with retries and better connection handling. Optimized feature flag and service initialization and streamlined security filter performance.

#### DevOps & Infrastructure

Improved smoke tests and diagnostics, fixed model schema definitions, and enhanced database migrations.

### May 2025

#### Application Enhancements

Expanded LLM provider support and improved universal endpoint compatibility. Added agentic workflows and enhanced Red Team insights with better visualization and application graphs.

#### Security & Compliance

Strengthened DLP handling, language validation, and security guardrails, including prompt injection and content filtering. Improved handling of security alerts and secret metadata.

#### Performance Improvements

Optimized provider API paths, improved endpoint reliability, and enhanced local DLP and context processing performance.

#### DevOps & Infrastructure

Improved CI/CD and Rust code quality and standardized Docker configurations.

### April 2025

#### Application Enhancements

Improved streaming handling and fallback support for Bedrock models. Enhanced guardrails APIs with richer tracing and metadata. Expanded security metrics to include PII categories and added support for content rejection headers. Released an updated Highflame Guard research paper with improved model evaluation and efficiency metrics.

#### Security & Compliance

Upgraded to Go 1.24.2 to address a critical security vulnerability. Fixed API key exposure in audit logs with proper masking. Improved model tracking and audit diffs for guardrail operations. Enhanced language detection for better threat prevention.

#### Performance Improvements

Improved model training efficiency with mixed-precision support and added memory usage tracking. Optimized guardrail routing and caching. Enhanced streaming parser performance and fixed a memory leak in prompt injection detection.

#### DevOps & Infrastructure

Improved smoke tests and database migrations for system reliability. Enhanced CI linting and validation. Strengthened ModelSpec configuration and dependency management. Fixed YARA scanning configuration in Ramparts.

### March 2025

#### Application Enhancements

Introduced an agent-based Red Team testing architecture. Simplified the API structure by removing legacy provider routes. Improved Redis high-availability handling, enhanced MCP metadata visibility, and updated DLP templates with better validation.

#### Security & Compliance

Improved prompt injection and code threat detection. Strengthened audit logging for API key operations. Added YARA-based scanning in Ramparts and fixed local DLP enforcement gaps.

#### Performance Improvements

improved query concurrency, reduced unnecessary API calls, and optimized DLP guardrail performance.

#### DevOps & Infrastructure

Improved CI and deployment validation.Added standardized modules, binaries, and Docker configurations.

### February 2025

#### Application Enhancements

Improved API routing reliability through better route name extraction. Enhanced ModelSpec handling for more accurate provider detection and base URL management. Simplified OpenAI client initialization and improved audit log visualization. Added language detection to strengthen content filtering.

#### Security & Compliance

Improved Guardrail handling for rejected model queries and strengthened model security checks, including reliable DLP enforcement.

#### Performance Improvements

Optimized error and context cancellation handling to reduce resource usage. Improved trace error visibility and efficiency of configuration update operations.

#### DevOps & Infrastructure

Upgraded core services to Go 1.24. Added YARA-based security scanning to Ramparts. Standardized API header handling.

### January 2025 <a href="#january-2025" id="january-2025"></a>

#### Application Enhancements <a href="#application-enhancements" id="application-enhancements"></a>

Introduced the concept of applications, allowing for structured app versioning and schema improvements. Added support for querying traces by trace ID, improving observability.

#### Security & Compliance <a href="#security--compliance" id="security--compliance"></a>

Introduced a new Threats table for storing security-related trace records. Improved KeyVault transaction handling and validation for better security.Implemented security enhancements to prevent CORS-related vulnerabilities.

#### Performance & Infrastructure <a href="#performance--infrastructure" id="performance--infrastructure"></a>

System Performance Metrics, Added avg\_model\_latency to system performance tracking. Enhanced Pub/Sub Handling, Reimplemented pub/sub for better Redis handling under load. Database Performance, Optimized query performance and improved indexing across key tables.

#### DevOps & CI/CD <a href="#devops--cicd" id="devops--cicd"></a>

Pipeline Enhancements, Improved CI/CD pipeline, including better PR status tracking and linting. Automated changelog updates as part of the release process. Integrated Trivy scans into the build pipeline for improved security monitoring.

### December 2024 <a href="#december-2024" id="december-2024"></a>

#### Security & Compliance <a href="#security--compliance-1" id="security--compliance-1"></a>

OpenAI-Compatible Model Specs, Updated model specs to align with Claude and OpenAI standards. Sensitive Data Protection, Enhanced filtering to prevent sensitive data leaks in API responses.

#### Performance & Optimization <a href="#performance--optimization" id="performance--optimization"></a>

Optimized Query Performance, Added new indexes to traces and threats tables. Enhanced Caching Mechanism, Improved caching system for API responses to reduce latency.

#### DevOps & Automation <a href="#devops--automation" id="devops--automation"></a>

Slack Plugin Updates, Bumped versions of Slack plugins in CI/CD pipelines. Version Bumps & Maintenance: Upgraded various dependencies and improved package management.

### November 2024 <a href="#november-2024" id="november-2024"></a>

#### New Features <a href="#new-features" id="new-features"></a>

Transformation Rules: Introduced transformation rules for model output handling. Advanced Gateway Filtering: Added gateway filtering by Business Unit ID, Organization ID, and Type.

#### Security & Compliance <a href="#security--compliance-2" id="security--compliance-2"></a>

Threat Table Enhancements: Added security detectors to trace records stored in the Threats table. Audit Logging: Improved audit logs to capture changes in routes, providers, and secrets.

#### Performance & DevOps <a href="#performance--devops" id="performance--devops"></a>

Upgraded Go Version: Migrated to Go 1.23.3 for performance improvements. Dockerfile Security Enhancements: Removed non-root users for improved container security.

### October 2024 <a href="#october-2024" id="october-2024"></a>

#### API & Infrastructure Improvements <a href="#api--infrastructure-improvements" id="api--infrastructure-improvements"></a>

Swagger Documentation Updates: Improved API documentation with better security schemes and operation IDs. KeyVault Enhancements: Enhanced transaction management and API key handling.

#### Security & Compliance <a href="#security--compliance-3" id="security--compliance-3"></a>

Regex & Keyword Filtering: Introduced regex-based filtering to prevent sensitive data leaks. Security Patch for CORS Handling: Improved CORS rules to prevent unauthorized API access.

#### Performance Enhancements <a href="#performance-enhancements" id="performance-enhancements"></a>

Optimized System Performance Metrics: Improved performance tracking for AI models. Improved CI/CD Pipelines: Enhanced build and deployment workflows with automated validation.

### September 2024 <a href="#september-2024" id="september-2024"></a>

#### Major Analytics & Telemetry Infrastructure Upgrade <a href="#major-analytics--telemetry-infrastructure-upgrade" id="major-analytics--telemetry-infrastructure-upgrade"></a>

Enhancements to the analytics infrastructure including pub-sub for large scale analytics capture and telemetry aggregations.

*Breaking change:* All analytics data will now be stored in a new data store. We are not migrating the old data to the new store so old data will not be available. Graphs, Analytics and Metrics will only show data from the date of the upgrade.

#### Regex & Keyword Filtering Support <a href="#regex--keyword-filtering-support" id="regex--keyword-filtering-support"></a>

We are introducing regex and keyword filtering support as a part of Highflame Data Protection. This will allow users to define custom filters based on regex patterns and keywords to protect sensitive data.

#### Enhanced Security Filters for Data Protection <a href="#enhanced-security-filters-for-data-protection" id="enhanced-security-filters-for-data-protection"></a>

Introduction of advanced security filters for code detection, markdown detection, and keyword/regex checks to prevent security vulnerabilities and data breaches.

#### Custom Processor Support <a href="#custom-processor-support" id="custom-processor-support"></a>

Highflame now supports a javascript-based custom processor that can be used to implement custom logic for API calls, including data transformations and response handling.

#### Bug Fixes & Enhancements <a href="#bug-fixes--enhancements" id="bug-fixes--enhancements"></a>

Various bug fixes and enhancements to improve the overall stability and performance of the platform.

### August 2024 <a href="#august-2024" id="august-2024"></a>

#### Security Dashboard Integration with OWASP LLM Top 10 <a href="#security-dashboard-integration-with-owasp-llm-top-10" id="security-dashboard-integration-with-owasp-llm-top-10"></a>

Launch of a comprehensive security dashboard categorizing vulnerabilities aligned with the latest OWASP Large Language Models (LLM) Top 10 risks, enhancing the ability to monitor and mitigate potential security threats effectively.

#### Comprehensive Audit Logging <a href="#comprehensive-audit-logging" id="comprehensive-audit-logging"></a>

Implementation of detailed audit logs for all user activities related to Routes, Providers, Data Protection, and Secrets management, ensuring traceability and compliance with security policies.

#### Enhanced API Call Tracing with OpenTelemetry <a href="#enhanced-api-call-tracing-with-opentelemetry" id="enhanced-api-call-tracing-with-opentelemetry"></a>

Introduction of OpenTelemetry-based tracing for all API interactions through the Gateway, allowing for precise performance monitoring and debugging capabilities.

#### Python SDK Enhancements <a href="#python-sdk-enhancements" id="python-sdk-enhancements"></a>

Upgrades and optimizations to the Python SDK to improve usability, performance, and compatibility with the latest Python versions.

#### Documentation Overhaul <a href="#documentation-overhaul" id="documentation-overhaul"></a>

Significant updates to API documentation, including revamped Swagger files and enhanced overall documentation, making it easier for developers to understand and integrate APIs.

#### Advanced Analytics Features <a href="#advanced-analytics-features" id="advanced-analytics-features"></a>

Enhancements to analytics functionalities, focusing on better data capture techniques and visualization tools to provide deeper insights into data usage and trends.

#### New Security Guardrails <a href="#new-security-guardrails" id="new-security-guardrails"></a>

Code/Markdown Detection: Implementation of advanced security filters designed to detect and handle code or markdown injections, ensuring content integrity and preventing security vulnerabilities.

### July 2024 <a href="#july-2024" id="july-2024"></a>

#### Roles and User Management Framework <a href="#roles-and-user-management-framework" id="roles-and-user-management-framework"></a>

Expanded support for managing roles, permissions, and user access across groups, teams, and entire organizations, enhancing control over user activities and data access.

#### Granular Privilege Control <a href="#granular-privilege-control" id="granular-privilege-control"></a>

Definition of distinct privileges for various user roles such as Super Admins, Super Members, Team Members, and Team Admins, ensuring appropriate access levels are maintained.

#### Permissions Enforcement on UI and API <a href="#permissions-enforcement-on-ui-and-api" id="permissions-enforcement-on-ui-and-api"></a>

Comprehensive permissions management across all UI pages and API calls, safeguarding against unauthorized actions and ensuring system security.

#### API Usage and Performance Analytics Dashboard <a href="#api-usage-and-performance-analytics-dashboard" id="api-usage-and-performance-analytics-dashboard"></a>

New analytics dashboard providing insights into API usage, performance statistics, error tracking, and latency measurements, aiding in efficient AI management.

#### Chronicle for Compliance <a href="#chronicle-for-compliance" id="chronicle-for-compliance"></a>

Detailed chronicles for all AI calls, including logs, error reports, and performance metrics specifically for Route Configuration, helping in troubleshooting and optimizations.

#### UI Enhancements for Navigation <a href="#ui-enhancements-for-navigation" id="ui-enhancements-for-navigation"></a>

Updates to UI components such as sidebars, headers, and footers, aimed at improving the overall user navigation experience.

#### Extension Processor for Custom API Logic <a href="#extension-processor-for-custom-api-logic" id="extension-processor-for-custom-api-logic"></a>

Introduction of an Extension Processor to implement custom logic for API calls, including data transformations and response handling, with support for gRPC.

#### New Security Guardrails and Streaming Support <a href="#new-security-guardrails-and-streaming-support" id="new-security-guardrails-and-streaming-support"></a>

Implementation of new security measures like keyword checks and code vulnerability scanners, alongside support for real-time data streaming technologies such as Redis and Kafka.

### June 2024 <a href="#june-2024" id="june-2024"></a>

#### Websocket Support in Core Platform <a href="#websocket-support-in-core-platform" id="websocket-support-in-core-platform"></a>

Integration of WebSocket technology into the core platform to enable real-time, bi-directional communication between clients and servers.

#### Support for more models and frameworks <a href="#support-for-more-models-and-frameworks" id="support-for-more-models-and-frameworks"></a>

Incorporation of support for non-OpenAI models, such as Bedrock, allowing for greater flexibility and diversity in machine learning operations.

#### Playground for Real-Time Testing <a href="#playground-for-real-time-testing" id="playground-for-real-time-testing"></a>

Establishment of a playground environment for real-time testing and experimentation with new features and configurations.

#### Enhanced Guardrails for Team and Global Settings <a href="#enhanced-guardrails-for-team-and-global-settings" id="enhanced-guardrails-for-team-and-global-settings"></a>

New functionalities for setting and managing security guardrails at both team and gateway levels, ensuring consistent security practices across all operational levels.

#### UI Improvements for Guardrail Configurations <a href="#ui-improvements-for-guardrail-configurations" id="ui-improvements-for-guardrail-configurations"></a>

User interface updates to facilitate the definition and management of guardrails, tailored to team and organizational requirements.

#### Configurations Across Teams and Organizations <a href="#configurations-across-teams-and-organizations" id="configurations-across-teams-and-organizations"></a>

Enhanced support for managing configurations across teams and organizations, providing scalability and flexibility in deployments.

#### Automation Enhancements for Private VPCs <a href="#automation-enhancements-for-private-vpcs" id="automation-enhancements-for-private-vpcs"></a>

Improved automation processes for private Virtual Private Cloud (VPC) deployments, optimizing resource utilization and deployment efficiency.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.highflame.ai/changelog/readme.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.