KamaClaude
youngyangyang04/KamaClaude
KamaClaude is a mini local implementation of a Claude Code AI agent system, designed to teach the inner workings of Agent loops, tool usage, memory management, and TUI interfaces.
196 88MITUpdated 2026-06-10
75
GreatOverview
KamaClaude is an educational project and local agent runtime built to demystify advanced AI programming tools like Claude Code and Codex. Divided into eight progressive stages ranging from basic IPC architecture and event streaming to tool security, context compaction, and MCP integration, it guides developers through constructing a fully functional daemon-client agent framework from scratch.
Capabilities
- ▸ReAct agent loop execution with multi-step planning
- ▸JSON-RPC 2.0 over NDJSON inter-process communication between core daemon and clients
- ▸Real-time event streaming and TUI rendering
- ▸Tool parameter validation, permission approval, and failure handling
- ▸Session, thread, and notes memory stratification
- ▸Context window monitoring, result truncation, and automatic/manual compaction
- ▸Support for custom skills, sub-agents, and Model Context Protocol (MCP)
Best for
Learning the internal architecture and engineering principles behind AI agents, Building a portfolio project for technical interviews demonstrating agentic workflows, Experimenting with local agent execution, tool registries, permission management, and context window governance
Works with
Claude CodeCursorChatGPT
Evoa Score breakdown= Σ (score × weight)
75Task usefulness20%100 → +20.0
実タスクにどれだけ役立つか（機能の豊富さ・用途の明確さ）。 — AIによるcapabilities/use-cases解析
Code quality15%85 → +12.8
実装・指示の品質。 — AIによるSKILL.md/README解析
Maintenance15%85 → +12.8
リポジトリがどれだけ活発に保守されているか。 — GitHub 最終push日時の新しさ
Documentation12%30 → +3.6
ドキュメントの充実度・分かりやすさ。 — README/独自要約の情報量
Security15%50 → +7.5
危険・不審な挙動が無いか。 — AIによるセキュリティレビュー
Originality10%90 → +9.0
ありふれたラッパーではない独自性。 — AIによる独自性判定
Popularity8%64 → +5.1
コミュニティの採用度。 — GitHub Stars/Forks（対数スケール）
Compatibility5%75 → +3.8
対応AIエージェントの広さ。 — AIによる対応エージェント判定
Weighted total74.5 / 100
ライセンス不明/制限あり（Red）のSkillは総合スコアに0.85倍の補正を適用します。 ランキングはこのScoreのみで決まり、広告で変わりません。 算出方法の詳細 →
Security considerations
The system includes built-in permission approval flows and tool parameter validation before executing local file or system operations.
Categories
Summary and analysis are original content generated by AI Skills Rank. The skill's source text is not reproduced here — view it on the linked repository.