Skip to main content

​
v1.3.4 — January 9, 2026

​
Faster insights, clearer observability, and smoother releases

​
Search and investigate server logs directly from agents

You can now let agents like Eve search your server logs on demand to troubleshoot issues without leaving Genesis.
  • New grep_logs tool exposes regex-based log search to agents with configurable context lines and log-level filtering
  • Supports searching across rotated log files and large log streams efficiently
  • Comes with comprehensive unit tests to ensure reliability over time
  • Opens the door to agent-driven incident investigation and health checks
(From: #2581)

​
Make metrics configuration visible and safe in the Config UI

Metrics are now fully managed through the unified configuration system, with safer naming and clearer options.
  • Renames the internal metrics.type setting to metrics.mode to avoid UI keyword conflicts
  • Adds dropdown-based options for metrics configuration in the Advanced Settings UI
  • Keeps documentation and config schema aligned so changes are discoverable and auditable
(From: #2578, #2467, #2466, #2439, #2516, #2530)

​
Safer, more predictable tool access

High‑risk tools are now locked down by default, with clearer policies and metrics for how your bots behave.
  • Permanently disables high‑risk tools like arbitrary Python execution and dbt from the default Eve toolset
  • Changes the default security posture so these tools cannot be turned on from configuration alone
  • Adds RBAC-controlled scheduler APIs and a new MANAGE_SCHEDULER permission for managing scheduled prompts
  • Improves logging and anonymized usage metrics so you can see how bots are used without exposing PII
(From: #2561, #2464, #2465)

​
Smarter authentication-aware UX across the app

The UI is now more resilient around authentication, reducing confusing errors and making sign‑out behavior consistent.
  • Adds per‑provider auth capabilities so the UI only shows logout where the provider supports it
  • Defers data fetching and WebSocket connections until authentication is fully ready
  • Improves log streaming pages with better colors, horizontal scroll, and clearer error handling
  • Refines thread and notification loading to avoid race conditions during page load
(From: #2577, #2505, #2496, #2493)

​
Dark mode and theming improvements

The Genesis UI now feels more modern and consistent, especially in dark environments.
  • Adds full dark‑mode support with theme tokens across core pages and components
  • Refines theming for buttons, inputs, logs, and sidebars in both light and dark modes
  • Updates status labels (for example, “Stuck” → “Awaiting Input”) to better reflect mission state
(From: #2423, #2471, #2486)

​
Faster, more scalable query results and dashboards

Working with large result sets is smoother, with server‑side pagination and more control over dashboard population.
  • Upgrades query result grids (both in the Data page and embedded in chat) to AG Grid server‑side row model
  • Enables server‑side pagination, filtering, sorting, grouping, and pivoting over arbitrary SQL queries
  • Adds per‑source maxRows configuration in dashboard manifests with a higher default limit (50,000 rows)
(From: #2492, #2328)

​
More efficient charting with query IDs

Charts generated by agents now stream much faster while using far fewer tokens.
  • Lets Plotly charts reference stored query results via data_source.query_id instead of embedding raw data
  • Adds a preview_rows parameter so agents only see a sample while the UI works with the full dataset
  • Improves streaming UX for charts by showing a dedicated “Streaming chart data…” placeholder during generation
(From: #2562, #2494)

​
Cleaner, more discoverable changelog and release workflows

The internal changelog and release pipelines have been simplified so GA promotions are more reliable.
  • Changelog generation now runs between semantic versions, GA promotions better understand RC artifacts, and Snowflake/Helm/Railway deploy flows are cleaner. This reduces release friction for teams shipping Genesis into multiple environments.
(From: #2564, #2499, #2500, #2501, #2556, #2557, #2552, #2553, #2554, #2460, #2461, #2463, #2447, #2454, #2457, #2480, #2479, #2504)

​
Helm and deployment flexibility for Databricks LLMs

Teams standardizing on Databricks for both data and AI can now deploy Genesis with Databricks LLMs more easily.
  • Adds Helm values for Databricks‑backed LLM and embedding models (Claude 4.5 family, GTE embeddings, etc.)
  • Supports multiple secrets strategies: direct values, existing Kubernetes secrets, or External Secrets operators
  • Auto‑detects which provider to use based on configured keys and sets appropriate environment variables
(From: #2542, #2473)

​
UI polish and documentation for configuration and knowledge

Configuration and knowledge management flows are cleaner and better documented.
  • Adds a comprehensive guide for adding new configuration fields end‑to‑end
  • Introduces a Config Sources API and UI views showing where each configuration value comes from
  • Removes the legacy Knowledge system and cleans up associated UI, routes, and jobs in preparation for a new approach
(From: #2476, #2485, #2570)

​
Preview / Beta Features

These capabilities are available today but may still evolve based on feedback.
  • Server‑side SSRM grids in more places – Query result SSRM is in active rollout; we will extend this pattern to additional data views over time. (From: #2492)
  • Agent‑driven log search – grep_logs is designed for agent use and will gain additional presets and UI affordances in upcoming releases. (From: #2581)

​
What This Means for You

🔐 Stronger security and governance High‑risk tools are now locked down by default, scheduler access is permissioned via RBAC, and metrics are anonymized and structured. This makes it easier to adopt Genesis in regulated environments while still observing how agents behave. 📊 Better observability and troubleshooting You can search logs directly via agents, see accurate source filenames in logs, and rely on richer metrics to understand usage. Combined with improved log UI and SSRM grids, investigating issues and exploring data is more efficient. 🚀 Smoother releases and deployments Changelogs are generated between semantic versions, GA promotions better understand RC artifacts, and Snowflake/Helm/Railway deploy flows are cleaner. This reduces release friction for teams shipping Genesis into multiple environments. 🌓 A more consistent user experience Dark mode, improved theming, authentication‑aware UX, and clearer status labels all contribute to a more polished daily experience for everyone working in the Genesis UI.

​
Under the Hood

  • Upgraded SQLAlchemy to 2.0 and refreshed several dependencies, including MCP server libraries and security‑sensitive packages
  • Added composite metrics services so the same metric can be logged and exported to SQL backends
  • Improved Bedrock and embedding provider normalization and proxy handling
  • Cleaned up legacy deployment scripts and workflows in favor of unified RC/GA and Snowflake deployments
  • Hardened CI workflows and security audits with clearer guidance and automated checks
(From: #2507, #2528, #2512, #2511, #2510, #2509, #2478, #2479, #2471, #2547)
If you have feedback on this release or ideas for what would make Genesis even more useful for your team, we’d love to hear from you. — The Genesis Team