Systematically tracking pain points across all major coding CLI agents to find what developers actually need — and contribute solutions.
Inspired by agents-radar, but laser-focused on coding CLI tools — the terminal-based AI coding assistants that are reshaping software development.
The coding CLI space has 300K+ open issues across 7 major projects. Real user pain points are buried in noise. This repo:
- Automated scanning — GitHub Actions weekly/monthly scans of issues, PRs, and releases
- Pain point extraction — Categorize and rank by community signal (reactions, duplicates, regressions)
- Cross-tool analysis — Find patterns that appear across 2+ tools (= universal developer needs)
- AI-powered analysis — Digests are analyzed via Kimi Code API for deeper insights
- Contribution roadmap — Turn insights into concrete PRs for projects that need them
| Project | Stars | Issues | Language | Core Focus | Target Users |
|---|---|---|---|---|---|
| OpenCode | 117K | 6,400+ | TypeScript | Model-neutral, Zen free tier, desktop | Price-sensitive, multi-model users |
| Gemini CLI | 97K | 2,700+ | TypeScript | Remote Agents, Tracker, Google ecosystem | Cloud-native developers |
| Claude Code | 75K | 5,600+ | TypeScript | Agent autonomy, Plan mode, context-aware | Pro devs wanting "autopilot" |
| OpenAI Codex | 63K | 1,600+ | Rust | Enterprise stability, plugin ecosystem | Enterprise teams, scalable arch |
| Qwen Code | 20K | 580+ | TypeScript | Chinese optimization, coding-plan, LSP | Chinese devs, Alibaba Cloud users |
| Copilot CLI | 9K | 990+ | Shell | GitHub workflow integration, compliance | GitHub power users, enterprise |
| Kimi CLI | 7K | 176 | Python | Plans project mgmt, OpenAI compat, Chinese | Chinese devs, multi-model mix |
| Project | Runtime | Architecture Highlights |
|---|---|---|
| Claude Code | Node.js/Bun | Deep Claude model binding; TUI-heavy interaction |
| OpenAI Codex | Rust | Full rewrite; Fast/Standard dual mode; OTEL observability |
| Gemini CLI | Node.js | Remote Agents infrastructure; Google Cloud integration |
| Copilot CLI | Rust + Electron | GitHub auth system; evolving from tool to agent infra |
| Kimi CLI | Python | Plans system → project management; AgentHooks security |
| OpenCode | Rust + TUI | Ghostty terminal; model-neutral; commercial model TBD |
| Qwen Code | Rust | Kitty CSI-u terminal protocol; i18n Insight reports |
| Category | Description | Affected Tools |
|---|---|---|
| Shell & Terminal | Command execution, CWD, shell integration | All |
| MCP Protocol | Tool loading, timeout, reconnection | All |
| Subagents | Lifecycle, cancellation, inter-agent | Claude, OpenCode, Kimi, Gemini |
| Context & Memory | Compaction, token limits, memory leaks | All |
| Plan Mode | Task planning, step-by-step, approval | All (various maturity) |
| Hooks & Plugins | Lifecycle events, extensibility | Claude, Kimi, OpenCode |
| Permissions & Security | Sandbox, file access, safety | All |
| Windows Compatibility | Path handling, PowerShell, EEXIST | All |
| IDE Integration | VS Code, JetBrains, Zed, Cursor | All |
| Cost & Model Routing | Token usage, model selection, quota | Claude, OpenCode, Codex |
| UX & TUI | Input, status line, multiline, a11y | All |
| Chinese/i18n | IME, CJK, localization | Claude, Kimi, Qwen |
Reports are generated in digests/ and also published as GitHub Issues:
- Weekly Digest — Top issues, trending categories, cross-references
- Monthly Deep Dive — Category-specific analysis with implementation recommendations
- Cross-Reference — Universal pain points found across multiple tools
- Contribution Opportunity — Concrete PR specs ready for implementation
GitHub Issues (7 repos) → scan-issues.sh → Raw Digest
↓
Kimi Code API Analysis
↓
Categorized + Ranked Report
↓
GitHub Issue + digests/ commit
The analysis step uses Kimi Code API to:
- Classify issues into pain point categories
- Identify cross-tool patterns automatically
- Score contribution opportunity value
- Generate actionable PR specs
We're contributing battle-tested solutions to Kimi CLI:
| Area | Kimi Issue | Reference | Priority |
|---|---|---|---|
| Plan Mode | #1354 | Claude Code + OpenCode implementations | High |
| Hooks System | #1313 | Claude Code PreToolUse/PostToolUse | High |
| Subagent Lifecycle | #1315 | Cross-tool patterns | Medium |
| Context Compaction | #1329 | Memory-efficient strategies | Medium |
| Windows Fixes | #1341 | 20+ Claude Code Windows fixes | Medium |
| MCP Reliability | #1316 | MCP regression patterns | Medium |
./scripts/scan-issues.sh # Last 7 days
./scripts/scan-issues.sh 30 # Last 30 daysWorkflows in .github/workflows/ run automatically:
- Weekly scan: Every Monday 09:00 UTC
- Monthly deep dive: 1st of each month
Every coding CLI tool faces the same fundamental challenges — but each solves them independently. By systematically tracking what breaks and what users want, we can:
- Accelerate development of younger projects with proven patterns
- Avoid known pitfalls documented across 300K+ issues
- Build a portfolio of high-impact open source contributions
- Map the landscape for anyone building or evaluating coding CLI tools
PRs welcome! You can help by:
- Adding new coding CLI tools to track
- Improving scanning scripts and AI analysis
- Writing cross-reference analyses
- Submitting PRs to tracked projects
MIT