GitHub AI-Agent Star Velocity

from pathlib import Path
import os

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd

from examples.hot_trends.data import (
    HotTrendDataError,
    append_real_snapshot,
    build_arxiv_monthly_counts,
    fetch_coingecko_market_chart,
    fetch_defillama_stablecoin_chains,
    fetch_github_repo_metadata,
    fetch_github_stargazers,
    fetch_huggingface_models,
    fetch_wikipedia_pageviews,
    source_audit_table,
)
from examples.hot_trends.decomposition import (
    component_summary,
    decompose_table,
    editorial_priority,
    residual_event_table,
)
from examples.hot_trends.scoring import article_publication_phrasing

pd.set_option("display.max_columns", 80)
pd.set_option("display.max_rows", 80)
plt.rcParams.update({"axes.grid": True})

CACHE_DIR = Path("examples/hot_trends/cache")
OUTPUT_DIR = Path("examples/hot_trends/outputs")
CACHE_DIR.mkdir(parents=True, exist_ok=True)
OUTPUT_DIR.mkdir(parents=True, exist_ok=True)

def save_table(df, name):
    path = OUTPUT_DIR / f"{name}.csv"
    df.to_csv(path, index=False)
    print(f"saved: {path.as_posix()}")

repos = [
    "langchain-ai/langchain",
    "microsoft/autogen",
    "crewAIInc/crewAI",
    "browser-use/browser-use",
    "modelcontextprotocol/servers",
]
token = os.environ.get("GITHUB_TOKEN")
pd.DataFrame({"repo": repos})

metadata_rows = []
for repo in repos:
    meta = fetch_github_repo_metadata(repo, token=token)
    metadata_rows.append({
        "repo": repo,
        "stars": meta.get("stargazers_count"),
        "forks": meta.get("forks_count"),
        "open_issues": meta.get("open_issues_count"),
        "pushed_at": meta.get("pushed_at"),
        "source": "GitHub REST API",
        "endpoint": f"https://api.github.com/repos/{repo}",
    })
metadata = pd.DataFrame(metadata_rows).sort_values("stars", ascending=False)
metadata

STARGAZER_PAGES = 3
PER_PAGE = 100
source_card = pd.DataFrame([{
    "source": "GitHub REST API",
    "endpoint": "https://api.github.com/repos/{repo}/stargazers",
    "access_date": pd.Timestamp.today().date().isoformat(),
    "query_params": f"repos={len(repos)}; pages={STARGAZER_PAGES}; per_page={PER_PAGE}; token_used={bool(token)}",
    "time_range": "derived from fetched starred_at timestamps below",
    "cache_path": "not cached; outputs saved to examples/hot_trends/outputs",
    "interpretation_scope": "star velocity measures developer attention in the fetched window; stars are not production usage",
}])
star_rows = []
for repo in repos:
    sg = fetch_github_stargazers(repo, pages=STARGAZER_PAGES, per_page=PER_PAGE, token=token)
    star_rows.append(sg)
stars = pd.concat(star_rows, ignore_index=True)
display(source_card)
stars.head(20)

stars["date"] = pd.to_datetime(stars["starred_at"]).dt.normalize()
raw_daily = stars.groupby(["repo", "date"]).size().reset_index(name="count")
calendar_rows = []
for repo, sub in raw_daily.groupby("repo"):
    calendar = pd.date_range(sub["date"].min(), sub["date"].max(), freq="D")
    filled = sub.set_index("date").reindex(calendar, fill_value=0).rename_axis("date").reset_index()
    filled["repo"] = repo
    calendar_rows.append(filled[["repo", "date", "count"]])
daily = pd.concat(calendar_rows, ignore_index=True)
daily["date"] = daily["date"].dt.date.astype(str)
raw_daily["date"] = raw_daily["date"].dt.date.astype(str)
daily.head(20)

star_velocity = daily.copy()
star_velocity["date"] = pd.to_datetime(star_velocity["date"])
velocity_pivot = star_velocity.pivot_table(index="date", columns="repo", values="count", aggfunc="sum").fillna(0).sort_index()
ax = velocity_pivot.rolling(7, min_periods=1).mean().plot(figsize=(11, 4), title="7-day average GitHub star velocity")
ax.set_ylabel("stars/day")
ax.set_xlabel("date")
plt.tight_layout()
plt.show()

image/png

coverage = source_audit_table(raw_daily, value_col="count", entity_col="repo", time_col="date")
calendar_coverage = daily.groupby("repo").agg(calendar_days=("date", "nunique"), total_fetched_stars=("count", "sum")).reset_index().rename(columns={"repo": "series"})
coverage = coverage.merge(calendar_coverage, on="series", how="left")
coverage["pages_per_repo"] = STARGAZER_PAGES
coverage["per_page"] = PER_PAGE
ready = coverage.loc[coverage["calendar_days"] >= 14, "series"].tolist()
if ready:
    components = decompose_table(daily[daily["repo"].isin(ready)], entity_col="repo", time_col="date", value_col="count", method="MA_BASELINE", period=7, trend_window=5, transform="log1p")
    summary = editorial_priority(component_summary(components, entity_col="repo", time_col="date"), entity_col="repo")
    events = residual_event_table(components, entity_col="repo", time_col="date", top_n=20, trim_edges=7)
else:
    components = pd.DataFrame()
    summary = pd.DataFrame([{"status": "not_enough_stargazer_calendar_days", "required": "increase pages or run repeated snapshots"}])
    events = pd.DataFrame()
coverage

coverage_plot = coverage.sort_values("observations")
fig, axes = plt.subplots(1, 2, figsize=(12, 4.5))
coverage_plot.plot(kind="barh", x="series", y="observations", ax=axes[0], color="tab:blue", legend=False, title="Observed stargazer days")
axes[0].axvline(14, color="tab:red", linestyle="--", linewidth=1.0)
axes[0].set_ylabel("")
if events.empty:
    axes[1].axis("off")
    axes[1].set_title("No residual events until coverage threshold is met")
else:
    event_plot = events.copy()
    event_plot["date"] = pd.to_datetime(event_plot["date"])
    repo_order = coverage_plot["series"].tolist()
    event_plot["repo_code"] = pd.Categorical(event_plot["repo"], categories=repo_order, ordered=True).codes
    sc = axes[1].scatter(event_plot["date"], event_plot["repo_code"], s=50 + event_plot["abs_residual_z"] * 35, c=event_plot["residual_z"], cmap="RdBu_r")
    axes[1].set_yticks(range(len(repo_order)))
    axes[1].set_yticklabels(repo_order)
    axes[1].set_title("Residual star-velocity events")
    fig.colorbar(sc, ax=axes[1], label="residual z")
plt.tight_layout()
plt.show()

image/png

summary

phrasing = article_publication_phrasing()
phrasing

save_table(source_card, "04_github_source_card")
save_table(metadata, "04_github_repo_metadata")
save_table(coverage, "04_github_stargazer_coverage")
save_table(daily, "04_github_star_velocity_daily")
save_table(summary, "04_github_decomposition_or_collection_status")
if not events.empty:
    save_table(events, "04_github_residual_events")
save_table(phrasing, "04_github_publication_phrasing")

saved: examples/hot_trends/outputs/04_github_source_card.csv
saved: examples/hot_trends/outputs/04_github_repo_metadata.csv
saved: examples/hot_trends/outputs/04_github_stargazer_coverage.csv
saved: examples/hot_trends/outputs/04_github_star_velocity_daily.csv
saved: examples/hot_trends/outputs/04_github_decomposition_or_collection_status.csv
saved: examples/hot_trends/outputs/04_github_residual_events.csv
saved: examples/hot_trends/outputs/04_github_publication_phrasing.csv