OSS Impact Analysis

Measure observable changes in open source software development patterns before and after widespread AI-assisted coding adoption (Oct 2022 - present), using GitHub API data from major repositories.

What This Skill Does

This skill helps you analyze GitHub repositories to detect changes in development patterns after AI coding tools became widespread. It collects commits, PRs, issues, and releases from multiple repositories, performs statistical analysis, and generates interactive visualizations.

Prerequisites

Instructions

Initial Setup

1. **Install dependencies**

```bash

uv sync --all-extras

```

2. **Configure pre-commit hooks**

```bash

uv run pre-commit install

uv run pre-commit install --hook-type commit-msg

```

3. **Set up environment variables**

- Create `.env.local` file

- Add `GITHUB_TOKEN=your_token_here`

- Optional overrides: `TEST_REPO`, `DATE_RANGE_START`, `DATE_RANGE_END`, `COLLECT_REPOS`

Data Collection Workflow

The project uses a **two-sweep collection** strategy:

**Sweep 1 - Git (Fast, No API Limits)**

```bash

uv run python scripts/collect_all.py --sweep git

```

**Sweep 2 - API (GraphQL, Rate-Limited)**

```bash

uv run python scripts/collect_all.py --sweep api

```

**Or run both sweeps together:**

```bash

uv run python scripts/collect_all.py --sweep all

```

Collection Options

```bash

Preview repositories without collecting

uv run python scripts/collect_all.py --dry-run

Force refresh all data (bypass cache)

uv run python scripts/collect_all.py --force-refresh

Collect specific repository only

uv run python scripts/collect_all.py --repo owner/repo

Re-collect stale repos (not updated in N days)

uv run python scripts/collect_all.py --stale-days 30

```

Adding New Repositories

1. Edit `src/config.py` and add to `CLUSTER_MAP`:

```python

CLUSTER_MAP = {

"owner/repo": "cluster-name",

}

```

2. Run collection for the new repo:

```bash

uv run python scripts/collect_all.py --repo owner/repo

```

3. Validate data quality:

```bash

uv run python scripts/validate_data.py

```

Data Validation

Run comprehensive data quality checks:

```bash

Standard validation

uv run python scripts/validate_data.py

Verbose mode (includes monthly gap details)

uv run python scripts/validate_data.py --verbose

```

Generate Analysis Summary

Create machine-readable findings:

```bash

uv run python scripts/generate_summary.py

```

This generates `output/findings_summary.json` with statistical results.

Interactive Dashboard

Launch the Streamlit dashboard for visual exploration:

```bash

uv run streamlit run dashboard/app.py

```

Dashboard pages:

Jupyter Notebook Analysis

For interactive exploration:

```bash

uv run jupyter notebook notebooks/01_explore_single_repo.ipynb

```

Available notebooks:

Code Quality

```bash

Lint code

uv run ruff check .

Format code

uv run ruff format .

Run tests

uv run pytest

```

Project Structure

```

src/

config.py # Settings, constants, repo clusters

models.py # Pydantic data models

storage.py # SQLite operations

collector.py # Collection orchestration

git_collector.py # Local git operations

graphql_collector.py # GitHub GraphQL API

charts.py # Plotly visualizations

metrics.py # Statistical analysis (35+ functions)

analytics/ # Dashboard analytics layer

data_service.py # Data loading, caching

repo_analytics.py # Single-repo analysis

cohort_analytics.py # Multi-repo aggregations

comparison.py # Repo vs repo comparisons

derived_metrics.py # Computed metrics

caching.py # LRU cache utilities

dashboard/

app.py # Streamlit entry point

pages/ # Dashboard pages

components/ # Reusable UI components

data/ # SQLite database (gitignored)

repos/ # Git clones (gitignored)

output/charts/ # Exported visualizations

```

Key Concepts

Data Model

Core tables:

Collection status flow:

```

pending → git_in_progress → git_complete → api_in_progress → completed

```

GitHub API Rate Limits

Estimated time for full cohort (21 repos):

Analytics Functions

The `src/analytics/` module provides UI-agnostic functions:

```python

from src.analytics import load_repo_data, get_repo_summary, compare_repos

Load repo data

repo_data = load_repo_data(repo_id)

summary = get_repo_summary(repo_data)

Compare repos

comparison = compare_repos(repo_a, repo_b, metrics=("commits", "prs_merged"))

```

Caching Strategy

Two-layer caching for performance:

Clear all analytics caches:

```python

from src.analytics.caching import clear_all_analytics_caches

clear_all_analytics_caches()

```

Important Notes

Output Files