Volcano Engine Serverless Flink Skill

volc-flink-skill 1.0.4 - Improved volc-flink-skill security and consistency by standardizing config path handling, removing risky credential-copy guidance, masking auth output, and excluding local auth directories from packaging.- Behavior, configuration, and routing logic remain unchanged from the previous version.

volc-flink-skill v1.0.3 - 新增 flink-cdc 子技能，支持 Flink CDC 任务、YAML/pipeline 模板、MySQL 多源同步等相关能力 - flink-sql 参考资料与模板集大幅扩充，覆盖更多数据源与典型场景 - 明确区分 CDC 任务（flink-cdc）与 SQL/JAR 任务（flink-sql、flink-jar）的路由规则 - 移除旧的 flink-template 相关模板与文档，统一最佳实践和模板输出入口 - 对路由冲突表和路由说明流程进行了细化、完善

- Declared required binaries - Added required_binaries to the entry skill and relevant subskills to explicitly require the volc_flink CLI. - Declared potential local config access - Added may_access_config_paths to document that the skill may read/write local Volc Flink configuration directories: - ~/.volc_flink - $VOLC_FLINK_CONFIG_DIR - Declared credential sources - Added a credentials section describing the primary credential source and optional environment variables commonly used in non-interactive flows: - VOLCENGINE_ACCESS_KEY - VOLCENGINE_SECRET_KEY - VOLCENGINE_REGION

### Breaking Changes - volc_flink jobs metrics now requires --sub-namespace : upgraded from “recommended” to mandatory , with runtime non-empty validation. - Metrics CLI guidance tightened : users must provide at least one -m/--metric ; if unsure about the namespace, they should run volc_flink jobs metrics list first. ### Added - Evaluation coverage expanded : introduced and registered multiple new eval suites/cases across SQL mutation, dependency management, catalog browsing, JAR drafts, savepoints, SRE flows, batch drafts, and dedicated performance/diagnostic metrics scenarios. - Formal evaluation report : added docs/report/iteration-2-eval-report.md including test time window, model (MiniMax 2.7), pass rate, latency, token/cost breakdown, and suite-level summary tables. ### Improved - flink-volc installation experience : switched to pipx-first installation guidance (venv as fallback), added guidance for root / externally-managed Python environments, and removed “install from local sf_cli directory” instructions. - flink-performance & flink-diagnostic metrics correctness : aligned metric usage with vcm_flink_metrics.md , reinforced --sub-namespace awareness, added a standard “metric not found” fallback procedure, and added Kafka metrics query examples. - flink-sql engine version robustness : improved Flink version recognition and mapping to --engine-version , and added regression coverage requiring explicit FLINK_VERSION_1_20 when Flink 1.20 is requested.

## 1.0.0 Initial release of Volcano Engine Flink Skill. ### Added - Unified entry skill for Volcano Engine Flink workflows - Intelligent routing across sub-skills for: - CLI install/login and account management - project and config management - catalog and resource pool operations - Kafka connection and message sampling - SQL and JAR draft creation - dependency management - savepoint and SRE operations - monitoring, diagnostics, and performance analysis - reusable Flink SQL templates and best practices ### Included Skills - `flink-volc` - `flink-auth` - `flink-config` - `flink-projects` - `flink-resource-pool` - `flink-catalog` - `flink-dependency` - `flink-conn` - `flink-sql` - `flink-validate` - `flink-jar` - `flink-template` - `flink-sre` - `flink-savepoint` - `flink-batch` - `flink-monitor` - `flink-diagnostic` - `flink-performance`