Skip to content

v1.2: Embedded Model Context Protocol (MCP) Server & Amazon SQS Support

Choose a tag to compare

@github-actions github-actions released this 11 Sep 18:32
· 3 commits to main since this release

Streamer v1.2 ⚡

Streamer v1.2 brings two major additions to the streaming suite: Amazon SQS queue management and an embedded Model Context Protocol (MCP) Server that enables AI coding assistants (Claude Desktop, Cursor, Antigravity) to safely discover, inspect, and query your streaming infrastructure directly from chat!


🤖 Embedded Model Context Protocol (MCP) Server

Streamer now runs a built-in local MCP SSE Server (http://127.0.0.1:<port>/sse), bridging modern AI assistants directly to your message brokers:

  • Token-Efficient 4-Meta-Tool Interface:
    • list_actions: Discover available operations across Kafka, NATS, and SQS filtered by protocol or read-only status.
    • get_action_schema: Fetch detailed JSON parameter schemas on demand.
    • run_action: Execute single streaming operations with intelligent parameter normalization and JSON payload serialization.
    • run_action_sequence: Execute multi-step streaming pipelines with output piping between steps.
  • Strict Safety Guarantee: No delete or purge actions are ever exposed to MCP, ensuring AI agents cannot accidentally destroy cluster data.
  • Read-Only Safety Switch: Toggle Read-Only mode in Settings to restrict AI agents exclusively to query and inspect operations.
  • 1-Click Integration: Ready-to-use configuration snippets exportable directly into claude_desktop_config.json or Cursor/Antigravity MCP settings.
  • Settings & Lifecycle: Toggle server on/off with live status pill, configure loopback port (default 8765), and optionally enable auto-start on application launch.

📬 Amazon SQS Suite

Full-featured AWS and local SQS queue exploration and telemetry:

  • Standard & FIFO Queues: Create, inspect, and configure Standard and FIFO (.fifo) queues with full parameter control (Visibility Timeout, Message Retention, Delivery Delay, Content-based Deduplication).
  • Live Message Polling: Long-polling inspector supporting Peek Mode (inspect messages without affecting visibility) and Consumer Mode.
  • Interactive Message Producer: Send messages with custom Delay Seconds, Message Group IDs, Deduplication IDs, and custom String/Binary Message Attributes.
  • Dead Letter Queue (DLQ) Redrive: Automatic detection of linked DLQs from Redrive Policies and safe batch redrive of dead-letter messages back to active source queues with progress tracking.
  • Local & Cloud Compatibility: Native support for AWS SQS, LocalStack, and ElasticMQ endpoints.

⚡ Connector Hardening & Bug Fixes

  • Kafka Empty Payload Fix: Fixed an issue where callers passing value, message, data, or raw JSON objects produced 0-byte Kafka records. All payloads are now safely extracted and serialized.
  • NATS KV Value Hardening: Audited and hardened nats.put_kv_entry to correctly serialize structured JSON objects and reject empty values.
  • SQS URL Resolution: Added automatic queue name-to-canonical URL resolution for SQS MCP actions.
  • Parameter Normalization: Added robust synonym and alias mapping across all Kafka, NATS, and SQS actions.

🛠️ Verification & Tech Stack

  • Pure Go (Zero CGO): Franz-Go (Kafka), Nats.go (NATS), AWS SDK v2 (SQS), Mark3labs (MCP), Modernc (SQLite).
  • Cross-Platform Binaries: Pre-compiled binaries for Windows (streamer-windows-amd64.exe) and Linux (streamer-linux-amd64).
  • Cryptographic Checksums: Full SHA256 integrity verification in SHA256SUMS.txt.

Full Changelog: v1.1.0...v1.2