# Summary: 2026-07-15_17-05-06Z_EarlyAdoptionofAgenticCodingToolsbyGitHubProjects.md
Saved: 2026-07-15 21:01
Source: 2026-07-15_17-05-06Z_EarlyAdoptionofAgenticCodingToolsbyGitHubProjects.md
Model: None

---

## Summary  
The paper investigates how open‑source projects adopt agentic coding tools that autonomously generate pull requests, focusing on project‑level adoption patterns and collaboration dynamics. It analyzes a large dataset of agent‑generated PRs to quantify productivity, participation rates, and oversight models. The study reveals that adoption is limited to a few repositories and varies by size, with human oversight largely centralized. These findings highlight early organizational challenges in integrating autonomous code contributions.  

## Key Contributions  
- Finding 1: Median repository generates only one to two agentic PRs over three months, indicating intensive adoption remains concentrated in a small subset of projects.  
- Finding 2: Small projects (1‑5 contributors) show higher participation ratios and average levels of agentic PR activity compared with medium‑size or large repositories.  
- Finding 3: Human‑agent collaboration is dominated by a single‑human oversight model, while multi‑human collaboration patterns are uncommon.  

## Methodology  
The authors collected data from 25,264 agentic pull requests submitted to 2,361 popular GitHub repositories over a three‑month period. They extracted repository metadata, contribution histories, and PR details, then performed descriptive statistics and cross‑sectional analyses to compare adoption rates across project sizes and compute productivity metrics.  

## Results  
The median repository generated one to two agentic PRs in the observation window, far below the industry benchmark of 36 PRs per participant. Small projects had participation ratios up to 40 % versus about 15 % for larger ones. Only a handful of repositories exceeded the benchmark, while most stayed under it. Collaboration patterns were overwhelmingly single‑human, with no significant evidence of multi‑human review or modification.  

## Significance  
These results underscore that successful integration of agentic coding tools depends on both technical capability and organizational processes such as oversight structure and project size. They provide empirical insight into early adoption challenges and suggest that future research should monitor longitudinal trends in how open‑source communities manage autonomous contributions.  

## Related Concepts  
- Agentic coding tools (autonomous PR generation)  
- Human‑agent collaboration models  
- Open‑source project governance  
- Pull request productivity metrics  
- Contribution participation ratios