featurebench-microbenchmark
Creates new FeatureBench microbenchmarks for YugabyteDB's benchbase regression pipelines. It turns a plain-language requirement into an approved plan, picks the category and next file name, writes the yugabyte YAML, derives the postgres / yb_colocated / yugabyte_range variants, and lints them. It stops once the lint-clean files are in config/yugabyte/regression_pipelines and does not run them, commit, push or open a PR. Use it whenever the user wants to add, design, extend or port a microbenchmark, featurebench workload, or regression_pipelines YAML. That includes performance tests for a DB feature or query shape (scans, joins, indexes, aggregates, order by, foreign keys, DDL, locking, writes, vector, throughput), even if FeatureBench is not named, e.g. "write a yaml that measures X" or "add a microbenchmark for Y".
Source facts
- Repository
- yugabyte/benchbase
- Last source activity
- September 30, 2026 at 13:20
- Detected SKILL.md language
- English
- Stars
- 1
- Forks
- 10
Install options
The review-first prompt is selected by default. You can switch to a direct command or download a local copy.
Review the source files
Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.
Showing SKILL.md
| name | featurebench-microbenchmark |
|---|---|
| description | Creates new FeatureBench microbenchmarks for YugabyteDB's benchbase regression pipelines. It turns a plain-language requirement into an approved plan, picks the category and next file name, writes the yugabyte YAML, derives the postgres / yb_colocated / yugabyte_range variants, and lints them. It stops once the lint-clean files are in config/yugabyte/regression_pipelines and does not run them, commit, push or open a PR. Use it whenever the user wants to add, design, extend or port a microbenchmark, featurebench workload, or regression_pipelines YAML. That includes performance tests for a DB feature or query shape (scans, joins, indexes, aggregates, order by, foreign keys, DDL, locking, writes, vector, throughput), even if FeatureBench is not named, e.g. "write a yaml that measures X" or "add a microbenchmark for Y". |