跳至正文
OpenAgentSkill
浏览已开放 · 目录记录日期 2026-11-21 · 自动更新和账户服务暂时关闭。
金融

oss-evaluate-repo

Evaluate whether an OSS repo is worth long-term investment before committing time. Assesses project health, governance, community, bus factor, and trajectory. Use when deciding whether to contribute to a specific repo, choosing between multiple repos, or evaluating a project's sustainability. Not for checking if a repo accepts contributions — use oss-find-issue for that.

chiruu12 · 已有静态检查记录

安装与使用

运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。

npx skills add chiruu12/OSS-Skills --skill oss-evaluate-repo

技能说明


name: oss-evaluate-repo description: | Evaluate whether an OSS repo is worth long-term investment before committing time. Assesses project health, governance, community, bus factor, and trajectory. Use when deciding whether to contribute to a specific repo, choosing between multiple repos, or evaluating a project's sustainability. Not for checking if a repo accepts contributions — use oss-find-issue for that.

Evaluate Repo

Is this repo worth your time? Not every open source project deserves months of contribution effort. This skill evaluates project health, governance, community dynamics, and trajectory — so you invest in repos that will value your work and still exist in a year.

Purpose

Contributing to OSS is an investment. A single meaningful PR takes 10-40 hours including learning the codebase. Before investing that time, you should know: Is this project actively maintained? Is the community healthy? Will your contributions be reviewed? Could the project be abandoned next month? This skill answers those questions with evidence, not vibes.

When to Use

  • Deciding whether to invest time in a specific OSS project
  • Choosing between multiple repos to contribute to
  • Evaluating a project before committing to a GSoC proposal
  • NOT for checking if a repo accepts external contributions — oss-find-issue does that
  • NOT for evaluating code quality — oss-explore-repo does that
  • NOT for repos you're already contributing to — too late, you're invested

Prerequisites

  • A repo URL or name to evaluate
  • gh CLI authenticated
  • Clear goals for why you want to contribute (learning, career, community)

Process

1. Check vital signs

These are binary health checks — if any fail, the repo is likely not worth investing in.

# Basic repo info
gh api repos/{owner}/{repo} --jq '{
  stars: .stargazers_count,
  forks: .forks_count,
  open_issues: .open_issues_count,
  archived: .archived,
  license: .license.spdx_id,
  created: .created_at,
  updated: .pushed_at,
  default_branch: .default_branch
}'

# Recent commit activity
git log --oneline -20 --since="6 months ago" 2>/dev/null || \
  gh api repos/{owner}/{repo}/commits --jq '.[0:10] | .[] | "\(.commit.author.date[:10]) \(.commit.message | split("\n")[0])"'

# Is the repo archived or in maintenance mode?
gh api repos/{owner}/{repo} --jq '.archived'

Kill signals (stop evaluation if any are true):

  • Archived or read-only
  • No commits in 6+ months
  • README says "no longer maintained" or "looking for maintainers"
  • License is missing or incompatible with your needs

2. Assess governance and bus factor

Who runs this project, and what happens if they leave?

# Top contributors (bus factor check)
gh api repos/{owner}/{repo}/contributors --jq '.[0:10] | .[] | "\(.contributions)\t\(.login)"'

# Recent committers (who's active NOW, not historically)
gh api repos/{owner}/{repo}/commits --jq '[.[0:50] | .[].author.login // "unknown"] | sort | group_by(.) | map({user: .[0], commits: length}) | sort_by(-.commits) | .[0:5] | .[] | "\(.commits)\t\(.user)"'

# Check if it's an org or personal project
gh api repos/{owner}/{repo} --jq '.owner.type'

Evaluate:

  • Bus factor: How many people have committed in the last 3 months? If it's 1, the project dies if they stop.
  • Org vs personal: Org-backed projects are more sustainable. Personal projects depend on one person's motivation.
  • Corporate backing: Check if contributors have @company emails or if the org is a company. Corporate-backed OSS has resources but may shift priorities.
  • Governance model: Is there a GOVERNANCE.md? A code of conduct? A clear decision-making process?

3. Evaluate community health

A healthy community means your contributions get reviewed, your questions get answered, and conflicts get resolved.

# Issue responsiveness — how quickly do issues get responses?
gh issue list -R {owner}/{repo} --state closed --limit 20 --json number,createdAt,closedAt,comments \
  --jq '.[] | {number, created: .createdAt[:10], closed: .closedAt[:10], comments: .comments}'

# PR review turnaround
gh pr list -R {owner}/{repo} --state merged --limit 20 --json number,createdAt,mergedAt,reviews \
  --jq '.[] | {number, created: .createdAt[:10], merged: .mergedAt[:10], reviews: (.reviews | length)}'

# Check for toxic interactions (look at recent closed issues/PRs)
gh issue list -R {owner}/{repo} --state closed --limit 5 --json number,comments \
  --jq '.[].number' | while read n; do
    echo "=== Issue #$n ==="
    gh issue view $n -R {owner}/{repo} --json comments --jq '.comments[-3:] | .[].body' | head -20
  done

Does an outsider actually get merged? This is the single question the rest of the section is circling, and it needs counting rather than eyeballing. Note that gh pr list has no authorAssociation field. Use the REST API:

gh api "repos/{owner}/{repo}/pulls?state=closed&per_page=100" \
  | jq -r '[.[] | select(.merged_at != null)]
           | "sample: \(length) merges out of the last 100 closed PRs",
             (group_by(.author_association)[] | "  \(length)\t\(.[0].author_association)")'

# Then the humans behind the outside slice. Bots are not evidence of anything
gh api "repos/{owner}/{repo}/pulls?state=closed&per_page=100" \
  --jq '.[] | select(.merged_at != null) | select(.user.type != "Bot")
        | select(.author_association != "MEMBER" and .author_association != "OWNER")
        | .user.login' | sort | uniq -c | sort -rn

Read the sample size before the split. That endpoint pages over closed PRs, and closed is not merged: on cli/cli the last 100 closed contain 63 merges, on pydantic-ai 46. Rejected PRs skew toward outsiders, so the merged slice is always smaller than the page and never a clean hundred. Quote the number you actually got.

author_association describes permissions on this repo, not employment, so it does not answer the question on its own. A

阅读完整来源 ↗