Skip to content

0.7.0

Choose a tag to compare

@github-actions github-actions released this 23 Sep 12:38
30f4928

✨ Highlights

🤖 Connect AI agents to OpenOps with the new MCP server

OpenOps now ships an MCP server that lets AI agents such as Claude Code or Codex help you create FinOps workflows. Connect an agent to your OpenOps environment and describe the cost optimization you want to automate. The agent explores your available blocks, connections, and existing workflows to propose a workflow design tailored to your setup.

The MCP server is opt-in. Set OPS_OAUTH_ENABLED=true, OPS_OAUTH_RS_CLIENT_SECRET, and COMPOSE_PROFILES=mcp in your .env to enable it. See the MCP server docs for details.

⚙️ Faster, more resilient workflow execution

The engine container has been replaced by a new openops-worker container that runs each workflow in its own isolated engine process. A warm pool of ready processes removes cold-start latency, per-process memory limits stop one workflow from affecting others. If you use the release Docker Compose file or the install script the switch is automatic; custom deployments should replace the openops-engine service with openops-worker and can drop OPS_ENGINE_URL. Pool size is tunable with OPS_ENGINE_POOL_MIN_SIZE, OPS_ENGINE_POOL_MAX_SIZE, and OPS_ENGINE_COLD_FORK_CONCURRENCY.

🚀 Features and Improvements

  • Added bulk actions for workflows in the workflows list: move to folder, export, and delete selected workflows
  • Added "Batch Add Records" and "Bulk Update Rows" actions to the OpenOps Tables block
  • Added bulk account roles support to AWS connections, so multiple account roles can be added at once
  • Added follow-up question support to the Slack and Microsoft Teams "Request Action" actions
  • Added a configurable Microsoft Tenant ID field to the Microsoft Teams connection
  • Added a "Vendor Account IDs" input to the Cloudability "Get Recommendations" action
  • Replaced the engine container with a new openops-worker container that runs workflows in a pool of isolated engine processes.
  • Added an allowPauseResponse option to synchronous webhooks so callers receive a response as soon as the workflow pauses
  • Updated the AI provider model lists with the latest models from Anthropic, OpenAI, Google, xAI, DeepSeek, Mistral, Cerebras, and others, and removed deprecated models
  • Added the Azure CLI bastion extension to the worker image
  • Added support for authenticating to AWS from AKS using workload identity federation

🐛 Bug Fixes

  • Fixed workflow duration being overcounted in nested action execution
  • Fixed a crash in text inputs with mentions when rendering the spec
  • Fixed step settings reloading when the browser window regains focus

🔧 Maintenance

  • Moved container images and the Helm chart from public.ecr.aws/openops to openops.azurecr.io
  • Upgraded the runtime to Node.js 24 LTS and updated Docker base images
  • Updated the OpenOps Tables image to 0.2.22 and the OpenOps Analytics image to 0.14.8
  • Fixed security vulnerabilities by bumping proxy-addr, nodemailer (8.0.11 → 9.0.1), SMTP/IMAP block dependencies, dompurify (3.4.10 → 3.4.13), js-yaml (5.2.0 → 5.4.1), nanoid (3.3.12 → 3.3.18), qs (6.15.2 → 6.16.0), and shell-quote (1.8.4 → 1.10.0)
  • Bumped fastify from 5.8.3 to 5.12.5, hono from 4.12.18 to 4.13.1, @hono/node-server from 1.19.14 to 2.1.0, typeorm from 0.3.30 to 0.3.31, and undici from 6.27.0 to 6.28.0
  • Bumped tiptap from 2.27.2 to 3.31.3, snowflake-sdk from 2.4.3 to 3.3.0, and i18next to 26.4.0
  • Upgraded the build toolchain to Nx 23, Vite 8, ESLint 9, and Jest 30

Contributors

Big thanks to

Full Changelog: 0.6.25...0.7.0