Release v1.6 · abstracta/tero

Another release with a new set of features and fixes! 🎉

Agents 🤖

New tools YouTrack and PractiTest integrations.
Add protected configuration for agents: global owners can mark an agent so others can use it in chat but cannot view, clone, or modify its settings.
Improve the agent import flow with clearer feedback when an import succeeds with adjustments — e.g. the model was replaced, some tools were skipped, or imported tools need authentication before use.
Refactor agent editor with collapsible sections and vendor logos.
Improve MCP, GitHub, and Redmine tool configuration with in-editor setup guidance — field helper text and placeholders (server URL, auth method, tokens), plus links to each service's official documentation.
Support for new models: Claude Sonnet 4.6, Claude Opus 4.6, GPT-5.4, and GPT-5.

You can now define tests with multiple interactions (user inputs and expected agent responses). Tests will fail fast comparing each agent response to the expected one.
Tests run in the background now, no longer blocking UI, interrupting tests or presenting some inconsistent behaviors when running and user closes tabs, tab deactivates, etc.
When running tests, proper authentication is requested if the agent uses any tool requiring it (like Jira or some MCP servers).

Test failures now show specific error codes with actionable guidance. For example, when the agent hits its step limit, the result explains what happened and suggests breaking the task into smaller parts.
Improve test runner stability and auto-generate descriptive names for test cases created from chat.
Fine-tune evaluation criteria per test case. Choose a different model, temperature, or custom prompt for specific tests that need stricter or more lenient grading.
Add chat export from conversations to share with colleagues or people without access to your Tero instance.

Improve extension visuals and auth flow for a more consistent and polished experience across the copilot and the main app.