Skip to content

Station 0.19.1

Choose a tag to compare

View all tags
@github-actions github-actions released this 17 Nov 05:35
· 295 commits to main since this release
v0.19.1
40ccd71

Station 0.19.1 🚂

Welcome to Station 0.19.1! This release includes improvements to the AI infrastructure platform that makes MCP servers as easy to discover, configure, and deploy as Docker containers.

Installation

One-liner install:

curl -sSL https://getstation.cloudshipai.com | bash

Or with specific version:

curl -sSL https://getstation.cloudshipai.com | bash -s -- --version v0.19.1

Manual download:
Download the appropriate binary for your platform from the assets below.

Go install:

go install github.com/cloudshipai/station/cmd/main@v0.19.1

Changelog

🚀 Features

  • 93ccbff: feat(evals): Add EntropyHelper for variable mock tool responses (@epuerta9)
  • 1479357: feat(evals): Update GuardDuty mock to use EntropyHelper (@epuerta9)
  • b9cc043: feat(faker): Add tool cache package for standalone faker mode (@epuerta9)
  • 7cfd55d: feat(faker): Implement standalone faker mode with AI tool generation (@epuerta9)
  • 9ae3121: feat(faker): add PRD multi-agent hierarchy scenarios (easy/medium/hard) (@epuerta9)
  • ca0e991: feat(faker): add complex AWS FinOps multi-faker workflow (@epuerta9)
  • d67e6cc: feat(faker): fix MCP timeout and validate complex multi-step executions (@epuerta9)
  • b754518: feat(faker): productionize faker with templates, API, and deterministic caching (@epuerta9)
  • 32b4e64: feat(finops): Add GCP FinOps standalone faker environment (@epuerta9)
  • be8009b: feat(otel): Add OpenTelemetry distributed tracing integration (@epuerta9)
  • 30f1902: feat(ui): Add hierarchical agent visualization with orchestrator and callable badges (@epuerta9)
  • 6a33bd8: feat(ui): Add right sidebar with scrollable agent list cards (@epuerta9)
  • 5fa577f: feat(ui): add execution flow visualization with Jaeger telemetry integration and selection highlights (@epuerta9)
  • e259a32: feat: Add FAKER_TOOL_NAMES env var for deterministic faker tool generation (@epuerta9)
  • ce58081: feat: Add MCP tools for model filtering in reports and runs (@epuerta9)
  • 4f2b6b9: feat: Add Turso/libsql database support for cloud-native persistence (@epuerta9)
  • eb739f8: feat: Add comprehensive Jaeger service tests and improve startup timeout (@epuerta9)
  • 8b81d56: feat: Add model filtering support for agent runs and reports (@epuerta9)
  • 92c0eee: feat: Enable telemetry by default and simplify Jaeger service to pure Docker (@epuerta9)
  • f0238fc: feat: Fix token tracking in scheduled runs and add auto-benchmark evaluation (@epuerta9)

🐛 Bug fixes

  • bb3f2eb: fix(faker): Add CLI support methods to session package (@epuerta9)
  • fb55434: fix(faker): enable standalone mode for agent execution (@epuerta9)
  • cd8db7f: fix(faker): resolve stn sync hanging with multiple AI-enabled fakers (@epuerta9)
  • eea4fc0: fix(faker): write all debug output to stderr for stdio MCP compatibility (@epuerta9)
  • 28f90be: fix(ui): Always show source handle on agent nodes for MCP connections (@epuerta9)
  • b020d68: fix(ui): Remove toolsApi call and use MCP server tools for hierarchy detection (@epuerta9)
  • fa0d1f3: fix(ui): Use toolsApi instead of direct apiClient call for hierarchy detection (@epuerta9)
  • 59891b8: fix: Add critical execution path bug fixes and environment validation (@epuerta9)
  • 55cb1db: fix: Dynamic Agent MCP port and enable sync on Fly.io startup (@epuerta9)
  • e1e3be0: fix: Fly.io configuration - use http_service for Dynamic Agent MCP on 8587 (@epuerta9)
  • 1b64113: fix: Make Jaeger startup fully async and use ephemeral storage (@epuerta9)
  • 3481d17: fix: Properly propagate agent execution errors instead of silently marking as completed (@epuerta9)
  • 40ccd71: fix: bind MCP servers to 0.0.0.0 for Fly.io deployment and restore registry pull (@epuerta9)

📚 Documentation

  • 73b0089: docs(faker): Add GenKit integration architecture and foundation (@epuerta9)
  • 68f578f: docs(faker): add E2E success summary for faker integration (@epuerta9)
  • 37a93ce: docs(faker): add comprehensive PRD for final faker testing phases (@epuerta9)
  • 6e04eca: docs(faker): comprehensive faker configuration and caching learnings (@epuerta9)
  • 61bb948: docs(faker): comprehensive session summary for Nov 9, 2025 (@epuerta9)
  • adc16cc: docs(otel): Add comprehensive OpenTelemetry setup guide and README section (@epuerta9)

🧪 Tests

Other

  • d38243e: Add Report MCP tools for environment-wide agent evaluation (@epuerta9)
  • 17623f2: Add STATION_AI_API_KEY to auto-init detection (@epuerta9)
  • e417b29: Add agent_agents table and repository for hierarchical agent relationships (@epuerta9)
  • 741ffcd: Add benchmark and schedule management MCP tools (@epuerta9)
  • ea2d97e: Add comprehensive GPT-4o vs GPT-4o-mini model comparison analysis (@epuerta9)
  • 3ee9a77: Add comprehensive deployment test results documentation (@epuerta9)
  • 99d31ff: Add debug logging for tool registration and discovery (@epuerta9)
  • 8940585: Add deployment DX improvements: one-line install and quick-deploy scripts (@epuerta9)
  • d8313c4: Add deterministic faker configuration examples (@epuerta9)
  • a8944bd: Add local base image build for Fly.io deployment (@epuerta9)
  • a8fa539: Add multi-agent hierarchy support via MCP (@epuerta9)
  • b79e4d9: Add orchestrator completion fix documentation (@epuerta9)
  • 348e1d8: Add remove_agent_as_tool MCP endpoint for managing multi-agent hierarchies (@epuerta9)
  • 8cccdcf: Add schedule variable schema validation to set_schedule MCP tool (@epuerta9)
  • 7c246eb: Clean up documentation and improve observability infrastructure (@epuerta9)
  • c70d92e: Fix Dockerfile to use tar extraction instead of stn bundle install (@epuerta9)
  • 0e1da9d: Fix Fly.io deployment path and build system refactor (@epuerta9)
  • 676cda9: Fix agent hierarchy runtime integration and UI visualization (@epuerta9)
  • aeada28: Fix agents: section export with proper newline formatting (@epuerta9)
  • a433a0e: Fix benchmark metrics API to include judge tokens, cost, and evaluation duration (@epuerta9)
  • 0d76c8d: Fix deployment AI provider configuration (@epuerta9)
  • 90f8cc8: Fix deployment script port handling and add comprehensive tests (@epuerta9)
  • ff633da: Fix docker-compose.yml generation and test with real Docker (@epuerta9)
  • 454a8d0: Fix faker standalone mode database migrations (@epuerta9)
  • 0d3240b: Fix multi-agent hierarchy infinite recursion by filtering agent tools properly (@epuerta9)
  • 56f1cb5: Fix orchestrator run completion bug with signal handling (@epuerta9)
  • 52aa596: Fix tool call counting by upgrading to GenKit v1.1.0 and using History() (@epuerta9)
  • 482a78f: Implement agents: frontmatter section for hierarchical agent relationships (@epuerta9)
  • 6527588: Improve report criteria descriptions and add error logging (@epuerta9)
  • 05a18c2: Improve zero-config Docker Compose example for better DX (@epuerta9)
  • ed58a55: Merge pull request #78 from cloudshipai/evals (@epuerta9)
  • f71a6ff: Refactor build system to use bundle-based deployment instead of baked databases (@epuerta9)
  • da9a8d9: Refactor: Extract AI package from monolithic faker (@epuerta9)
  • a6d5792: Refactor: Extract session package from monolithic faker (@epuerta9)
  • 32de104: Refactor: Extract telemetry package from monolithic faker (@epuerta9)
  • a7a01ba: Remove model comparison analysis from root directory (@epuerta9)
  • e8ef7b1: Simplify auto-init: Remove STATION_AUTO_INIT flag requirement (@epuerta9)
  • 5497ea9: Wire up SchedulerService to MCP Server lifecycle (@epuerta9)
  • 4dc5987: Wire up benchmark evaluation with MCP tools (@epuerta9)
  • aae38b1: feat(multi-agent): Add hierarchical agent execution with parent-child run tracking (@epuerta9)
  • ba384b5: proof(faker): empirical evidence of tool generation non-determinism (@epuerta9)
  • 6147a62: refactor(ui): Modularize AgentsCanvas into separate components (@epuerta9)
  • 3d518c8: remove: Eliminate STATION_SKIP_SYNC_ON_STARTUP environment variable (@epuerta9)

What's Station?

Station transforms complex MCP server setup from hours of configuration into a 30-second guided wizard. Think of it as the "package manager for AI agents."

Key Features:

  • 🧙 AI-powered MCP server discovery from GitHub URLs
  • 📋 Interactive configuration wizards with validation
  • 🚀 One-command deployment from repo to running server
  • 🤖 Revolutionary AI agent platform with smart context management
  • 🔒 Enterprise security with local and remote deployment modes

Getting Started

# Initialize Station
stn init

# Discover and deploy any MCP server from GitHub
stn load https://github.com/modelcontextprotocol/servers/tree/main/src/filesystem

# Access admin interface
ssh admin@localhost -p 2222

Full Changelog: v0.18.2...v0.19.1

Contributors

epuerta9
Assets 8
Loading