GitHub user maple525866 edited a discussion: Provide and continue to improve the relevant CLI and skills for Seata.
Hi everyone. I've been following the development of our Seata, and recently I noticed we have a benchmark CLI command-line tool. I think the current trend is towards CLI-based tools while weakening MCP. Our company also supports CLI for all basic services, along with corresponding skills and agents that can directly execute commands to perform actions. The biggest advantage of this is a completely closed-loop development process. Also, people unfamiliar with Seata can easily learn Seata operations using agents like CC and Codex. Therefore, I'd like to suggest that our community follow suit. Perhaps we could do some research on this. Thank you. I have the following suggestions: - Continue to improve the current Seata benchmark-cli tool. - Create a Seata diagnostic CLI, similar to: `seata-doctor check --server 127.0.0.1:8091 --registry nacos --config file`. This could check: server connectivity, registry/config readability, transaction grouping mapping existence, store mode configuration correctness, and DB/Redis connection. This is valuable for new users and troubleshooting online issues. - Configure the configuration center initialization CLI. Currently, `script/config-center/*` contains shell scripts for nacos, zk, apollo, consul, and etcd3. This is well-suited for unifying into a single Java CLI, such as: `seata-config init --type nacos --server-addr ... --namespace ...`. This is most valuable because it offers better cross-platform compatibility, working on both Windows and Linux, and reusing Seata's own configuration model. - Provide corresponding CLI tools and all necessary Seata-related skills. I'd be happy to contribute to this. Below is my initial architecture concept and design: The overall architecture consists of four layers: Layer 0 — Callers: Developer, AI Agent (Claude/Codex), and CI/CD Pipeline are the three entry points, all treated equally as command-line calls by the CLI. Layer 1 — CLI Entry Layer: Cyan = Existing seata-benchmark Purple = Suggested additions seata-doctor and seata-config Coral = Agent-oriented skills/MCP wrapper layer Layer 2 — Functional Modules: Each CLI is further subdivided into 2-3 single-responsibility modules, which is also the natural boundary for implementation. Note that the three inspection dimensions of seata-doctor (connectivity → vgroupMapping → store mode) are progressive, and the diagnostic order should also follow this progression. Layer 3 — Seata Core SDK: The core components Registry, Config (ConfigurationFactory), TM/RM/TC, and Store share the same set across all CLIs, avoiding redundant implementation of connection logic. This is the biggest engineering advantage of this proposal. <img width="1440" height="1440" alt="image" src="https://github.com/user-attachments/assets/533d7467-6df4-4eec-8e5c-0450a027b046"; /> GitHub link: https://github.com/apache/incubator-seata/discussions/8073 ---- This is an automatically sent email for [email protected]. To unsubscribe, please send an email to: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
