Release v0.4.0

Security & Query Safety

• Ensured only single statements are passed to EXPLAIN (#224), with improved parsing to block multiple commands.
• Renamed explainQuery to unsafeExplainQuery to highlight potential risks where edge cases might escape safety checks.
• Introduced safeExplainQuery, which fetches SQL directly from pg_stat_statements by queryId.
• Switched unsafe explain to use generic_plan for consistency and safety (#229).

AI Model Updates

• Migrated OpenAI integration to GPT-5 only (#211):

  • GPT-5 is now the default (chat + scheduling).
  • Removed support for GPT-4 and other deprecated OpenAI models.
  • Migration ensures schedules using GPT-4 are upgraded to GPT-5.
  • Pending LiteLLM and judge/eval testing.

Dev Experience & Tooling

• Removed unused sample MCP server (#218).
• Updated docker-compose behavior:
  • Containers now use unless-stopped restart policy for more predictable restarts across host reboots (#195).
  • Updated release references in docker-compose files.

Release v0.3.1

🛠️ Custom Tools via local MCP servers

• You can now add your own custom tools by creating an MCP server, which can be called locally (for now).
• Added UI pages for viewing and managing MCP servers.

🤖 Ollama Integration

• Added support for Ollama as a local LLM provider.

🧠 Reasoning Models

• Added support for reasoning models O1 and O4-mini.
• Adjusted tool parameters to be fully required for compatibility.

♻️ Refactored Components

• Replaced internal duplicated UI code with shared Xata components.
• Simplifies maintenance and promotes consistency.

📈 Monitoring Overhaul

• Switched to OpenTelemetry NodeSDK with Langfuse integration.
• Removed reliance on @vercel/otel, improving compatibility.

---

🐛 Bug Fixes

• OpenAI model provider is no longer mandatory. The agent falls back to another key if, for example, you only have the Anthropic key configured. PR: #186
• Fixed issue with tool prompts referencing a non-existent tool name. PR: #173
• Resolved dbaccess bug affecting scheduled jobs. PR: #172
• Removed misleading thresholds from the monitoring playbook (e.g., 20GB). PR: #166
• Persist and load the connection for chats. PR: #182

---

🙌 Thanks to Our Contributors

• Alexis Rico (@SferaDev)
• Tudor Golubenco (@tudor)
• Elizabet Oliveira (@miukimiu)
• Steffen Siering (@steffen)
• PineappleChild (@PineappleChild)

Release v0.2.2

This release includes fixes, quality-of-life improvements, and dependency updates.

• GCP Database Selector (#151)
  Fixed a bug where the database selector was incorrectly using IDs instead of names. Fixes #148.
• Database Selector on Schedule Edit Page (#149)
• Monitors Not Running (#147)
  Resolved an issue preventing monitors from executing as expected.
• Improved Built-in Provider Logic (#137)
• Model Column in Schedules Table (#150)
  A new column showing the model associated with each schedule was added, making it easier to identify models in multi-model projects.

Release v0.2.1

Patch release:

• Removing usage of crypto.randomUUID to fix self-hosting custom playbooks (#132)
• Initial LiteLLM support (#63)

Release v0.2.0

🚀 Features & Improvements

• 💬 Chat Enhancements

  • Redesigned chat UI with improved usability and visual polish (#52)
  • Sidenav improvements: rename & delete actions, refactored logic (#103)
  • Unified schema updates for new chat: history, votes, artifacts (#97)
  • Added support for new chat parameters (#121, #112)

• 🌐 Cloud Integrations

  • ☁️ Added GCP CloudSQL support, including metrics, logs, and onboarding improvements (#77)

• 🧩 Custom Playbooks

  • ✨ Support for creating, editing, and running custom playbooks via UI (#61)

• 🧠 Model Upgrades

  • Updated built-in models and changed the default to gpt-4.1 (#122)
  • Refactored model and provider logic for better ID handling (#98)
  • Introduced Gemini model support via Google Generative AI API (#82)

• ⚙️ Tooling Improvements

  • Added cluster/instance-aware tools for enhanced prompt context (#118, #73)
  • Tools can now retrieve metrics from the writer instance (not whole cluster) with min/max/avg/sum support (#113, #76)
  • Improved log handling for RDS instances (#105, #75)
  • Returned errors from tools in a model-friendly format instead of throwing (#91)

• 🔧 Developer Experience

  • Introduced Agent playground powered by Mastra (#56)
  • Refactored internal structure of tool modules (#62, #94)
  • Model Provider registry added for dynamic model listing (#89)
  • Better type safety for DB schemas (#87)

🛠 Fixes & maintainance

• 🐛 Fixed UPSERT and DELETE logic in artifact API (#104)
• 🐞 Corrected ID usage in model calls (#99)
• 🛠 Snapshot update after merge conflict (#100)
• 🧪 Fixed error when validating invalid URLs (#59)
• 🔌 Reused DB connections and added concurrency limits for DB tools (#67)

• 📦 Multiple dependency updates via automated PRs (#60, #86, #96, #102, #78, etc.)
• 🛡️ Don't expose internal database ports in Docker Compose (#74)

📘 Docs & Misc

• 📝 Updated roadmap/status in README (#84)
• 🧪 Initial evals test suite using PostgreSQL in Docker (#38)

🙌 Thanks to Our Contributors

• @leonardobenedet – Gemini model integration
• @PineappleChild – Custom Playbook support
• @SferaDev – Chat UI improvements, schema changes, model updates, artifact fixes, refactors, and more
• @tsg – GCP CloudSQL support, cluster tools, system prompts, documentation
• @elizabetdev – Co-authored UI updates, chat enhancements, and UX
• @urso – Playground, DB access abstraction, log and metric improvements, connection pooling
• @richardgill – Eval test infrastructure

Initial public release

This was released 13-03-2025, publishing the tag/release now for posterity.