Claude Memory Kit
技能 已验证 活跃Persistent memory for Claude Code agents with an agent-audit-ritual architecture. User only talks; agent captures, audits, proposes promotions, and writes. Four layers — daily logs, hot cache (MEMORY.md), role-based reference skills (Anthropic-native, user-invocable: false), canonical rules. /close-day runs the audit ritual at end of day. Multi-project isolation via projects/<name>/. Zero external dependencies.
To provide Claude Code agents with persistent, long-term memory that captures, organizes, and synthesizes information across sessions, enabling continuity and deeper contextual understanding.
功能
- Persistent memory across sessions
- Agent-driven information capture, audit, and promotion
- Four-layer memory architecture
- Multi-project isolation
- End-of-day audit ritual via /close-day
使用场景
- Maintaining brand voice and project context across multiple conversations
- Building a searchable knowledge base from agent interactions
- Ensuring continuity on complex, long-running projects
- Automating the synthesis and organization of daily work logs
非目标
- Direct user editing of memory files
- External dependencies or services
- Background detection scripts or auto-flushing
- Ad-hoc experimentation directly in production projects
安装
npx skills add awrshift/claude-starter-kit通过 npx 运行 Vercel skills CLI(skills.sh)— 需要本地安装 Node.js,以及至少一个兼容 skills 的智能体(Claude Code、Cursor、Codex 等)。前提是仓库遵循 agentskills.io 格式。
质量评分
已验证类似扩展
Wrap Up Ritual
100End-of-session ritual that audits changes, runs quality checks, captures learnings, and produces a session summary. Use when saying "wrap up", "done for the day", "finish coding", or ending a coding session.
Context Mode
100从 GitHub 更新 context-mode 并修复 hooks/settings。 拉取最新代码，构建，安装，更新 npm 全局包，配置 hooks。 触发器：/context-mode:ctx-upgrade
Ctx Doctor
100运行 context-mode 诊断。检查运行时、钩子、FTS5、 插件注册、npm 和 marketplace 版本。 触发器：/context-mode:ctx-doctor
Filesystem Context
98This skill should be used when the user asks to "offload context to files", "implement dynamic context discovery", "use filesystem for agent memory", "reduce context window bloat", or mentions file-based context management, tool output persistence, agent scratch pads, or just-in-time context loading.
Validate Plugin
100Validate a Claude Code plugin structure, frontmatter, and MCP tool references
Running Claude Code Via Litellm Copilot
100当通过本地 LiteLLM 代理将 Claude Code 路由到 GitHub Copilot 时使用，以减少直接的 Anthropic 支出，配置 ANTHROPIC_BASE_URL 或 ANTHROPIC_MODEL 覆盖，或对 Copilot 代理设置失败进行故障排除，例如 model-not-found、无 localhost 流量或 GitHub 401/403 身份验证错误。