Coding Agent Scorecard — AI skill for Claude Code
Compares how AI coding agents (Claude Code, Codex, OpenCode) handle their GitHub repos: issue backlog, closes, PRs, stars and npm downloads, plus agent-signed PRs across GitHub.
How to install Coding Agent Scorecard
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open akaiserg/coding-agent-scorecard and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Coding Agent Scorecard does
Compares how AI coding agents (Claude Code, Codex, OpenCode) handle their GitHub repos: issue backlog, closes, PRs, stars and npm downloads, plus agent-signed PRs across GitHub. Every number checked against GitHub's own counts. Python, standard library only.
Alternatives in AI
- Kilocode — GitHub Repo stars Open Source AI coding assistant for planning, building, and fixing code Open source AI assis 17k ★
- Context Engineering Intro — GitHub Repo stars Context engineering - the new way of AI coding assistants Context engineering guide 12.8k ★
- Codecompanion.nvim — GitHub Repo stars AI-powered coding seamlessly in Neovim Neovim AI coding 6.3k ★
README
AI coding harnesses on GitHub
How do the teams behind AI coding agents keep up with their own GitHub repositories? This project downloads every issue and pull request of three agent harnesses, measures them area by area, checks the numbers against GitHub's own counts, and turns them into reports, a dashboard and images for social posts.
| Key | Repository | Product source |
|---|---|---|
claude-code |
anthropics/claude-code | Private. The repository holds the issue tracker, plugins and examples. |
codex |
openai/codex | Public |
opencode |
anomalyco/opencode | Public |
It also measures the agents at work elsewhere: pull requests signed by each agent in any public repository on GitHub.
Snapshot: 27 Oct 2026, last 30 days
| Claude Code | Codex | OpenCode | |
|---|---|---|---|
| GitHub stars | 153,575 | 131,257 | 222,277 |
| New issues | 7,732 | 7,308 | 3,921 |
| Issues closed | 5,214 | 1,408 | 6,110 |
| Closed for every 100 new | 67 | 19 | 156 |
| Open issues at the end | 15,139 (+20%) | 24,907 (+31%) | 2,544 (−46%) |
| PRs merged | 13 ¹ | 1,270 ² | 1,611 |
¹ Public repository only. The product code is private, so its PRs say little about delivery. ² 99% are copies synced from an internal repository by the `copyberry` account.
"Closed" counts every close, including duplicates and issues closed as not planned. All the numbers above matched GitHub's search counts within 1%.
Agent-signed PRs in other people's repositories, created 20 Sep – 19 Oct 2026:
| Claude Code | Codex | OpenCode | Unsigned PRs | |
|---|---|---|---|---|
| Signed PRs found | 7.1M | 1.2M | 4.2k | 15.7M |
| Merged | 96% | 92% | 84% | 83% |
| Median lines changed | 229 |
Related Skills
Garl Receipt Action
Cryptographic verification for AI-authored code. ECDSA-signed receipts (secp256k1, RFC 6979 deterministic) for
Gh Issue ROI
Analyze repo issues without PRs, classify by difficulty + token/business impact. Args: owner/repo (e.g. rtk-ai
5dive
Run a company of AI agents on a server you own. Spin up named agents (claude, codex, pi…), put them on an org
Engage.Scorecard
Calibrate model verdict trust (Wilson-bounded miss-rate) to short-circuit re-validation
Review Video
Review a rendered video with AI vision. Uses TwelveLabs to watch the output, analyze pacing/composition/audio
Gestao De Projetos PM IA BSC DashBoard
🧭 Stop paying for AI in the dark. Turn ChatGPT/Claude/Copilot spend into auditable ROI: LLM observability → 7
Related Agents
Agent Maker
Creates new AI agent files in .claude/agents/ following AI Agents Convention. Changes are then synced to .open
Ecosystem Maintenance
Periodic health sweep across all Zanix repos for outdated/deprecated third-party dependencies (Deno std, npm p
Get Commit ID
gh api repos/OWNER/REPO/pulls/PR_NUMBER --jq '.head.sha' gh api repos/OWNER/REPO/pulls/PR_NUMBER/comments \ --