Roadmap

dbbot is currently an automated operations tool for the MySQL ecosystem, focusing on standardizing, documenting, and replayable deterministic database actions. The goal of Roadmap is not to be a chatting shell, but to gradually precipitate the underlying scripts, variables, validation and approval capabilities into a reusable operating system.

1. Current stage: dbbot for MySQL

The core capability of the current version is still dbbot for MySQL:

  • Use Playbook to solidify processes such as standalone, primary-replica, MGR, InnoDB Cluster, backup, recovery, and monitoring access.
  • Reduce operational drift with unified inventory, variable files and documentation.
  • Use standardized acceptance methods to ensure that successful deployment is not just “the task is completed”, but the results are verifiable.

The focus of this stage is to turn high-frequency operations actions related to the MySQL ecosystem into a stable, auditable, and reproducible infrastructure.

2. Next stage: Precipitate high-frequency actions into Skills

On top of Playbook, dbbot will continue to abstract common operations actions into smaller and more stable skills. Every skill should have:

  • Explicit input: topology, instance ports, variables, boundary conditions.
  • Clear output: deployment results, status inspection, and acceptance conclusions.
  • Clear pre-validation: version, host group, path, permissions, dependent packages.
  • Clarify failure handling: rollback, retry, remediation suggestions.

Actions suitable for precipitation into skills include:

  • Create a new MySQL instance or primary-replica topology
  • Install exporter for existing instances and register for monitoring
  • Perform backup, restore and post-restore data reconciliation
  • Perform standardized inspections for InnoDB Cluster or ClickHouse clusters

The goal of this step is to reduce the “human brain memory steps” and convert operational knowledge into composable and callable capability units.

3. Future stage: AI Agent natural language operation

After the skills foundation is stable, dbbot will introduce an AI agent entrance in the future, but this entrance will not directly replace the underlying Playbook, but will serve as a controlled orchestration layer.

The expected way to work is:

  1. The user describes the goal in natural language, such as “deploy a set of 8.4 MGR for 192.168.199.131/132/133 and access monitoring.”
  2. The AI ​​agent parses the intent into specific skills and parameters.
  3. The system generates execution plans, variable drafts and risk tips.
  4. After the user confirms, call the dbbot underlying Playbook for execution.
  5. After the execution is completed, the validation results, key outputs and supplementary suggestions are returned.

The key principles here are:

  • AI is responsible for understanding intentions and generating plans.
  • dbbot is responsible for performing controlled, verifiable low-level actions.
  • High-risk changes must retain manual confirmation and audit information.

4. Why do Skills first and then talk about Agent?

If the underlying action itself is unstable, the variables are unclear, and the acceptance criteria are inconsistent, then natural language will only hide the problem deeper. Precipitating skills first has three direct values:

  • Eliminates the need for the AI agent to directly assemble a large number of low-level parameters.
  • Allow the same action to be shared by document, command line, web and agent.
  • Let future natural language capabilities be built on verifiable and auditable execution units.

5. Current external commitments

In the current public version, AI agent natural language operation dbbot for MySQL still belongs to roadmap and should not be understood as a function that has been officially delivered. The current formal capabilities are still based on the Playbook, variable conventions and operating procedures disclosed on the documentation site.

If you are going to start using dbbot today, it is recommended to understand it in the following order:

  1. First use the Playbook and variable system of dbbot for MySQL.
  2. Then integrate supporting capabilities such as ClickHouse and Prometheus into the unified delivery process.
  3. Finally, consider encapsulating these standard actions into higher-level skills and agent entries.