Key features

Complete and intuitive observability of communications between components

• Capture communications from outside the applications
• Automatic higher level protocol analysis:
  • HTTP/1.x and HTTP/2 - both framings parsed the same way, on the same ports, with no configuration
  • HTTP/3 - HTTP over QUIC, decrypted from the captured TLS keys, with gRPC and SSE over HTTP/3, into the same HTTP communications
  • gRPC
  • PostgreSQL
  • Redis
  • Kafka
  • MySQL - text and prepared statements, bound parameters, result rows and query analysis, with TLS decrypted
  • MongoDB - one communication per command, replies paired exactly, query shapes, cursors followed across batches, with TLS decrypted
  • WebSocket and Server-Sent Events - a live stream becomes one communication per message, in both directions, over HTTP/1.1 and HTTP/2 - and SSE over HTTP/3
  • GraphQL - every call named by its operation and its real outcome, subscriptions included, on top of the HTTP communications
  • MCP (Model Context Protocol) - every call named by its method and tool, with its real outcome, answers streamed over SSE included
  • DNS - one communication per question and its answer, over UDP, TCP, DNS over TLS and DNS over QUIC, with retries and unanswered queries made visible
  • MQTT - MQTT 3.1.1 and 5.0: every publish with its acknowledgement, every delivery to a subscriber, every connection and subscription, with QoS handshakes, MQTT 5 properties, and payloads decoded from JSON, Protobuf, Avro, CBOR and Sparkplug B
• Unified multi-protocol view of HTTP/1.x, HTTP/2, HTTP/3, GraphQL, MCP, gRPC, PostgreSQL, MySQL, Redis, Kafka, DNS, MQTT, MongoDB and stream-message communications together
• Kubernetes workload identity on every TCP session, packet and communication - the namespace, kind, name and node of both endpoints, resolved as of capture time rather than guessed from a hostname
• Automatic correlation of one request across services - traceparent, X-Request-ID and X-Correlation-ID are extracted on HTTP, gRPC, Kafka, PostgreSQL, MySQL and MQTT 5 with nothing to configure
• Make binary payloads readable: transcoding of Protobuf and MessagePack bodies to searchable JSON via uploaded schemas
• Automatic communication deciphering: TLS, and QUIC for DNS over QUIC and HTTP/3
• TLS deciphering also for runtimes with no hookable TLS library (rustls, static BoringSSL; Envoy/Istio expected, not yet validated), through the opt-in key log fallback
• On Kubernetes - Complete observability of the cluster network usage

Visual, real time and dynamic analysis tools

• Ultra-fast and configurable statistical view of API endpoints usage
• Configurable dashboard to see issues at a glance
• Self construct and editable network maps and sequence diagrams
• Waterfall view to follow request timing and dependencies at a glance
• Topic map of MQTT, Redis pub/sub and Kafka - who publishes to which topic and who receives it, with QoS downgrades, silent subscriptions and orphan topics flagged
• Browse and analyse :
  • Network Hosts
  • HTTP/1.x and HTTP/2 communications, with their GraphQL and MCP tabs
  • gRPC communications
  • PostgreSQL communications
  • Redis communications
  • Kafka communications
  • MySQL communications
  • DNS communications
  • MQTT communications
  • WebSocket and SSE stream messages
  • Kubernetes workloads
  • TCP sessions, and UDP flows and conversations
  • Network packets
• Time travel from days to milliseconds
• Full-blown search with lots of helpers and filters
• Test your parsing rules live with the Whisperer config Playground - tags and templates validated against the real parsing engine
• Turn any search into a file: CSV export of communications, built server-side, with binary payloads decoded into readable columns
• Feed your own stack: RED and network-usage metrics over OTLP to the Prometheus and Grafana you already run
• Parse and analyse even you custom TCP protocols with plugins

No instrumentation

• No specific hardware required - amd64 and arm64 alike, Graviton and Ampere included
• No software dependency
• No langage dependency
• No need to update / hack your system
• Compatible with legacy and any off-the-shelf software

Easy to use

• Automated setup
• Remote deployment and management of agents
• Remote configuration
• Visual tools
• Command-line tools and AI-agent plugins to query captured traffic and self-monitoring outside the UI
• MCP server - paste one URL into Claude, ChatGPT or Claude Code, approve a consent screen, and the agent queries your captured traffic with your own rights, no API key to distribute
• OAuth 2.1 for AI agents - agents register themselves or identify with a Client ID Metadata Document, each connection is a revocable grant, and administrators choose which agent platforms may connect
• Monitoring and alerting embedded
• On premise solution