Complete Observability Into your AI Agents
Visualize agent workflow graphs with full observability, trace parameter flows across agentic assets, and gain clear insights into agent behavior.
Agentic Discovery
RAG Discovery
Discover agents using RAG automatically for better visibility into retrieval workflows.
Agentic Discovery
Agentic Risk Assessment
Understand the risk associated with each agent tool, MCP method, and operation with real-time risk scoring and classification.
Agentic Discovery
Agentic Inventory Insights
Get a high-level summary of your top Agents and MCPs for faster inventory assessment.
Agentic Discovery
Traces & Spans Observability
Traces and spans are now visible directly in agentic workflow details for full request-level observability.
Agentic Guardrails
Session-Based Guardrails
Apply guardrail policies consistently across the entire agent session, not just individual requests, for persistent policy enforcement.
Agentic Guardrails
Guardrail Playground
Run and validate your guardrail policies in the playground before deploying to production.
Agentic Guardrails
PII Types Synced with Sensitive Data
Users can now see PII types in guardrail configuration that reflect the data types defined on the Sensitive Data page.
Agentic Guardrails
Severity in Guardrail Policies
Set the severity level while creating a guardrail policy for better prioritization of security alerts.
Agentic Guardrails
Export Guardrail Activity to Webhook
Export the guardrail activity report to a custom webhook for SIEM or alerting pipeline integration.
Agentic Guardrails
Guardrail Controls
Enable or disable guardrails for specific agent tools and MCP methods, with automatic discovery of all agentic workflows through the agent proxy.
AI Red Teaming
Token Estimation for Scans
View token estimation for each Red Teaming scan before running it, so you can plan usage and estimate costs.
AI Red Teaming
Max Agent Token Limit
Set a maximum agent token limit while configuring a scan, enabling better control over usage and resource consumption.
AI Red Teaming
Scan Results from Previously Deleted Agents
Access scan results even after agents have been deleted.
AI Red Teaming
Select Probe Engine for Agentic Scans
Run and validate your agentic scans from the editor playground with support for probe engine and modifying hosts.
AI Red Teaming
View Probe Requests During Scans
Users can now see the actual probe requests sent during scans for full debugging visibility.
AI Red Teaming
Bulk Red Teaming Across Agentic Assets
Run bulk Red teaming across multiple AI Agents and MCPs, and view consolidated results in a single report.
Integrations
Vertex AI Integration
Discover and secure AI agents running on Google Vertex AI.
Integrations
TrueFoundry Integration
Monitor and secure AI agents deployed on TrueFoundry's MLOps platform.
Integrations
Arcade Integration
Secure your Arcade AI agent tool calls with Akto guardrails.
Integrations
QWEN Model Support (Ollama)
Use Ollama with confidence. Akto now supports the Qwen model in Ollama, expanding model coverage.
Reporting
Export Dashboard Tables as CSV
Export individual tables from Akto dashboards as CSV files for reporting and analysis.
Agentic Discovery
SkillGuard in Akto Atlas
Discover all agent skills installed on employee endpoints and analyze associated .md files for potential malicious instructions.
Agentic Discovery
RAG Discovery
Discover agents using RAG automatically for better visibility into retrieval workflows.
Agentic Discovery
Agentic Inventory Insights
Get a high-level summary of your top Agents and MCPs for faster inventory assessment.
Agentic Guardrails
Session-Based Guardrails
Apply guardrail policies consistently across the entire agent session, not just individual requests, for persistent policy enforcement.
Agentic Guardrails
PII Types Synced with Sensitive Data
Users can now see PII types in guardrail configuration that reflect the data types defined on the Sensitive Data page.
Agentic Guardrails
Guardrail Activity with User Attribution
Users can now see the associated username in guardrail activity on endpoints for better traceability.
Agentic Guardrails
Guardrail Playground
Run and validate your guardrail policies in the playground before deploying to production.
Agentic Guardrails
Severity in Guardrail Policies
Set the severity level while creating a guardrail policy for better prioritization of security alerts.
Agentic Guardrails
Export Guardrail Activity to Webhook
Export the guardrail activity report to a custom webhook for SIEM or alerting pipeline integration.
Agentic Guardrails
Guardrail Controls
Enable or disable guardrails for specific agent tools and MCP methods, with automatic discovery of all agentic workflows through the agent proxy.
Integrations
GitHub Copilot CLI Hooks
Monitor prompts and requests sent to GitHub Copilot CLI and enforce security guardrails in real time. Also supports VS Code agent workflows.
Integrations
Gemini CLI Hooks
Monitor prompts and requests sent to Gemini CLI and enforce security guardrails in real time.
Reporting
Export Dashboard Tables as CSV
Export individual tables from Akto dashboards as CSV files for reporting and analysis.