Skill based Agent architecture
Skills are reusable, task-focused building blocks that guide AI behavior in DataStoria. They help the assistant follow consistent workflows, apply ClickHouse-specific knowledge, and deliver reliable results for common tasks.
Benefits of Skills
- Token efficiency: Skills compress repeated instructions into a single, reusable unit.
- Consistent outcomes: Standardized steps reduce variability across similar tasks.
- Domain accuracy: ClickHouse-specific guidance improves query and diagnostic quality.
- Faster onboarding: Teams can share proven workflows without rewriting prompts.
- Safer automation: Skills can enforce guardrails and best-practice constraints.
ClickHouse Official Skills Supported
ClickHouse maintains an official Skills collection designed for AI assistants working with ClickHouse. These Skills provide high-level, domain-aware workflows for tasks like query exploration, performance tuning, and operational guidance, without requiring users to craft detailed prompts each time.
You can review the official catalog at:
- ClickHouse Agent Skills: https://github.com/ClickHouse/agent-skills
Use Skills in DataStoria
The agent automatically applies skills based on users' requests.
For example, if you ask such question:
Visualize the number of commits in 2021 Feb by day in line chartThe agent will automatically loads the Visualization skill while other skills like Optimization will not be loaded.
If you want to apply the official SKILLs, you can add 'best practice' keyword in your question to activate such skill. For example:
Apply the best practice to review the table: default.sampel_tableNext Step
To enable AI features, complete your model setup: