AI research agents are creating a verification economy
An open-source and scientific-publication cohort measures whether autonomous research systems produce reproducible artifacts, execute experiments, and expose independent verification.
Early evidence; the market is not yet confirmed
96 subject-filtered implementations show that autonomous research is productizing, but verification remains uneven. Code execution appears in 11.5% of the cohort, reproducible artifacts in 6.3%, experiment tracking in 3.1%, and explicit external verification in 1%. Research generation is advancing faster than independent validation.
Market snapshot
Comparable measurements from independent market surfaces.
Implementations and packages
Deduplicated and subject-filtered primary cohort.
Median repository stars
Calculated across repositories with at least one star.
With reproducible artifacts
6 items in the subject-filtered cohort.
With external verification
1 items in the subject-filtered cohort.
Search-match dynamics
Bars show monthly GitHub matches for AI-scientist and research-agent implementations. HF Daily Papers and other publications provide independent validation, not inflated repository counts.
What exists inside the category
One item may contain more than one feature.
Experiment code execution
11 · 11.5%
Reproducible artifacts
6 · 6.3%
Benchmarks and evaluation
9 · 9.4%
External verification
1 · 1%
Open research code
5 · 5.2%
Experiment tracking
3 · 3.1%
Physical or clinical validation
1 · 1%
Representative projects
Cross-source validation
These publications are not part of the primary numeric cohort.