research: task runner usage on most popular repos

[metaist]

I want to look at how people actually use task runners in the wild. This was a great post: https://aleyan.com/blog/2025-task-runners-census/

But I also want to look at specific features and see how much usage they get. I'm willing to aggregate features across runners to get a sense of what's useful.

A first approximation would be the top 100k GitHub repos (by stars). I'll have @claude suggest more particulars.

[metaist]

This comment was generated by Claude Opus 4.5 based on discussion with @metaist.

Research Plan

Scope

Repos to analyze:

• Top 200 repos by stars per language
• Languages: Python, JavaScript/TypeScript, Rust, PHP, Go, Ruby, Java, C/C++, C#, Swift (top 10)
• Estimated total: ~2000 repos

Task runner files to detect:

Runner	Files
Make	Makefile, GNUmakefile, makefile
npm/yarn/pnpm	package.json
Python (various)	pyproject.toml
Cargo	Cargo.toml
Composer	composer.json
ds	ds.toml
uv	uv.toml
Just	Justfile, justfile, .justfile
Task	Taskfile.yml, Taskfile.yaml
Deno	deno.json, deno.jsonc
Rake	Rakefile
Gradle	build.gradle, build.gradle.kts
Maven	pom.xml

Features to track

Core features:

• Composite tasks (dependencies/chains)
• Environment variables (inline)
• Working directory (cwd)
• Parallel execution
• Keep going / error suppression
• Help/descriptions

Advanced features:

• Argument interpolation ($1, $@, etc.)
• Environment files (.env loading)
• Pre/post hooks
• Workspaces/monorepo support

ds-specific:

• Verbatim mode
• Disabled tasks (# prefix)
• Shared options (_ task)

Implementation

Approach: GitHub API code search (no cloning needed)

Script location: research/task_runner_census/

Output:

• data/repos.json - list of repos analyzed
• data/results.json - raw feature detection results
• data/summary.csv - aggregated stats
• REPORT.md - human-readable findings

Execution: Batch with resume capability (saves progress, can continue if interrupted)

Goals (in order)

1. Compatibility check - ensure ds can handle common patterns
2. Feature prioritization - decide what to add/improve
3. Blog post - publish findings about task runner usage in the wild