npx @buildinternet/releases get dash0Recently shipped shared Agent0 skills, LLM-judged Evals, rebuilt traces, and an east-coast US region.
Agent0 skills became shareable and importable.
SKILL.md files from public GitHub/GitLab or private GitHub and refresh them with one click.Evals score agent quality as spans arrive. Evals are in beta and free during beta. Choose a trigger, span filters, sample rate, and a numeric, boolean, or categorical verdict; verdicts appear in the span's Evals tab and are stored as log records for dashboards and check rules.
Traces and AI SDLC Insights got faster and broader.
Navigation and billing got calmer and more transparent.
Protection and operations.
container.image.tag alongside container.image.tags until v0.165.0; update queries, alerts, and dashboards before then.--agent-mode take precedence over DASH0_AGENT_MODE; experimental datasets commands require --experimental.Failed checks can be silenced without changing the check rule, with silences tracked, extended, and ended in the Silences Hub for up to 7 days. Failed Check Notification automations do not run while a check is silenced, Agent0 can list, propose, and end silences, and a new silence can take up to 3 minutes to take effect.
Custom skills can be shared with members, teams, or the Admin and Member roles at Read or Edit, with each recipient enabling it via "Enabled for you" on the Skills tab; all members can now create skills, and the tab shows each skill's creator. Automations now use only explicitly allowed custom skills, and existing automations start with none until skills are added under Tools & Skills.
--agent-mode now takes precedence over DASH0_AGENT_MODE, so --agent-mode=false disables agent mode and all settings follow CLI flags, then environment variables, then profile, then default. New experimental datasets commands and Dash0Dataset support in apply arrive (requiring --experimental), alongside --config-dir and --auth-token-file global flags and fixes for deleted check rules, assets, and unquoted YAML y/n keys.
Billing & Plans now breaks down an organization's Agent0 credits by source, consumer, model level, and dataset, with filters and per-row cost links to manage each member, automation, eval, or token. If an Agent0 cap is set, it shows usage and projected end-of-period spend; data covers credits spent from October 5, 2026.
Dashboards and saved views can now be moved into folders by dragging them in the list, with dropping on the All dashboards header (or Custom header in views lists) removing them from folders. Built-in and externally managed items such as Terraform-managed ones cannot be moved, and the Move to dialog remains available in the menu.
Evals score agent generations as spans arrive using an LLM as the judge, with configurable trigger, span filters, sample rate, and numeric, boolean, or categorical verdicts. Verdicts appear in the span's Evals tab and are stored as log records for dashboards and check rules. In beta and free during the beta.
The SLO CRD now accepts bad/total and raw indicators, requiring rawType when raw is set, and the operator can manage its own webhook certificates. Fixed an api-sync bug where relist events starved the third-party sync queue, preserved the kubectl rollout restart annotation when updating workloads, excluded profiles from namespaced routing, and returned conversion errors for views and notification channels as 400.
Opening any span now streams the full trace instantly even at tens of thousands of spans, with search by name, filtering by type or errors, and logs and span events shown directly on the timeline.
AI SDLC Insights now has a Pull requests tab listing every pull request across repositories with the AI sessions behind it, their cost, and time to merge, plus a heatmap for selecting outliers, a status filter, a per-PR AI session side panel, Explore links from charts, and Stats totals.
The container.image.tag resource attribute is deprecated and will stop being emitted after operator v0.165.0, leaving only container.image.tags; queries, alerts, and dashboards using the old attribute need updating. Also fixed excessive HTTP requests during bulk resync and a nil check in the Prometheus metrics transform.
AI SDLC Insights now includes coding sessions from GitHub Copilot App and OpenCode in cost tracking, developer adoption metrics, and AI-assisted pull request analysis. Install the Dash0 plugin for GitHub Copilot App or add @dash0/agent-plugin-opencode-v2 to OpenCode, then run /dash0-configure.
Synthetic monitoring can now run from private locations inside your network, allowing checks against internal services with the same results and notifications. Each private location shows whether it is connected, and multiple locations can be run for resilience.
Adds OpenSLO CRD support, cluster-wide filters and transforms for operator configuration, and opt-in shared k8s_attributes processing between pipelines. The edge-proxy pprof option is now enabled on demand rather than by default, and reconcile triggers arriving mid-reconciliation are no longer dropped.
AI SDLC Insights now shows recommended actions based on coding-agent spend, adoption, and pull-request flow metrics. The Overview tab displays two to four Risk, Watch, or Opportunity findings with specific actions, supporting data on click, and thumbs-up/down feedback, refreshing only when numbers change.
Agent0 now reads data from AI SDLC Insights and returns answers matching the numbers on the page, aware of the current tab, time range, filters, and percentile settings. Each tab offers three suggested questions, and opening a session adds three more contextual questions.
Any page can now be pinned, moving it to a Pinned section at the top. Hovering a collapsed section opens its pages without expanding it.
Agent0 threads continue running when you switch away from them, and a new bottom bar lists running and recently finished threads with spinners for in-progress work and dots for finished ones. Keyboard shortcuts (Cmd+Shift+9 on Mac, Ctrl+Shift+9 on Windows/Linux) navigate between threads.
Automations are now editable only by admins and the members or teams they are shared with, where previously anyone with read access to a dataset could change or delete every automation in it. Settings now lists shared automations on each team and member page under a new Automations tab.
The Dash0 app has a new layout with content on its own panel, darker and calmer navigation, and most separator lines removed, using spacing and surface levels instead of lines to group elements. Nothing has moved, so no relearning is needed.