This server enables structured data collection and analysis for AI-assisted development, turning "vibe coding" into a measurable, improvable process that compounds knowledge over time.
Core Capabilities:
Track tasks (aida_task_start / aida_task_done): Log task start and completion with automatic duration calculation, grouped by module/stage.
Monitor bugs (aida_log_bug / aida_bug_fix): Record bugs with severity (critical/high/medium/low) and discovery source, then mark them resolved with fix notes.
Code reviews (aida_log_review): Document self-review results with pass/fail status, file scope, and issues found.
AI deviation tracking (aida_log_deviation): Capture instances where AI output didn't meet expectations, categorized by type (UI, API, logic, etc.) and root cause (hallucination, rule-missing, context-insufficient, etc.).
Automated file change logging (aida_log_files): Scan git diff to record modified files and line counts — no manual input required.
Highlights (aida_highlight): Record notable achievements like performance improvements or architectural wins.
Current status (aida_status): Get a real-time snapshot of the active development run — tasks, bugs, and overall progress.
Rule sedimentation (aida_log_rule): Convert recurring deviations into persistent project-level rules (component usage, API conventions, code style) that AI tools reference in future sessions.
Outcome: Each session builds on previous learnings, enabling data-driven retrospectives, sprint reviews, and a standardized SOP from requirement analysis through code generation, review, and rule sedimentation.
Provides specialized configuration and support for the Windsurf IDE, allowing for the collection and visualization of AI development metrics and performance patterns within that environment.
扫描 .claude/rules/、.cursor/rules/、.codex/、.aida/ 等目录，找出重复、矛盾、失效的规则和技能。所有写操作走 aida_apply_governance MCP 工具，每批 journal 一次，可通过 aida_undo 一键回滚。
核心 skill： /aida-govern（一站式）
用 MADR 格式把 "为什么这段代码这么写" 记录到 .claude/rules/decisions/。Claude Code 通过 paths 前置元数据在相关文件打开时自动加载，无需手动召回。
核心 skill： /aida-remember、/aida-remember-branch
本地 Dashboard（aida dashboard）提供 package.json 风格的资产清单：规则/技能/决策/MCP 配置/插件，一屏看完。
核心 skill： /aida-audit
Related MCP server: flaiwheel
/plugin marketplace add LWTlong/aida
/plugin install aida@aida重启 Claude Code。12 个 aida-* skill 会立刻可用。
升级：
/plugin update aida在项目根 .mcp.json 加入：
{ "mcpServers": { "aida": { "command": "npx", "args": ["-y", "aida", "mcp"] } } }或用一键脚本：
npx aida init类别 | Skill | 用途 |
治理 |
| 一站式：扫描 → 分析 → 确认 → 一次落盘（覆盖规则过载/任务型规则/文档冗余/跨工具重复/冲突/聚合镜像/2.x 遗留物） |
| 解决 AI 资产文件的 git 合并冲突 | |
| 跨工具同步：Claude ↔ Cursor ↔ Codex（你选源→目标） | |
| 从本地目录路径安全导入外部 Claude Plugin（先审计风险） | |
只读 |
| 扫描并梳理项目所有 AI 资产 |
记忆 |
| 记一条决策为 MADR，Claude Code 自动加载 |
| 从当前分支的 diff 批量提取决策 | |
| 召回之前沉淀的模块记忆 | |
| 分支/功能开发完后沉淀"改了什么、为什么" | |
| 回滚最近一次 AIDA 写操作 | |
分享 |
| 打包资产为 Claude Plugin |
| 打开本地 Dashboard（检查 npm 包 → 引导安装 → 给出启动命令） | |
| 显示 skill 参考卡片 |
工具 | 作用 |
| 初始化 / 状态检查 / 工具授权清单 |
| 扫描所有 AI 资产 |
| 浏览资产 |
| 执行治理动作（create/modify/delete/remove-lines），全部 journaled |
| 撤销上一次治理操作 |
| 写分析报告 |
| 读写决策记忆 |
| 打包项目资产为 Claude Plugin |
| 打包 AIDA 自身为 Plugin |
| 解析/审计外部 Plugin |
.claude-plugin/
plugin.json # Claude Plugin 清单（本仓库本身就是 plugin）
marketplace.json # Marketplace 清单
src/
assets/skills/ # 12 个 aida-* skill 源（plugin skills 字段指过来）
core/ # 扫描、治理、undo、memory、plugin builder
mcp/server.ts # MCP server
cli/ # aida CLI
dashboard/ # Dashboard 前端产物
.claude/
rules/decisions/ # MADR 决策记忆（Claude Code 自动加载）
.aida/
cache/ # 扫描结果 + undo journalnpm test
npm run buildUnclaimed servers have limited discoverability.
If you are the server author, to access and configure the admin panel.
Secure, user-owned long-term memory for AI agents over OAuth-protected remote MCP. Save, search, recall, update, and govern preferences, project context, decisions, and task state across ChatGPT, Claude, Copilot, IDEs, and CLIs.
Persistent memory and cross-session learning for AI coding assistants (hosted remote MCP).
One shared context your team's AI tools read & write over MCP. No re-explaining. Free.
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/LWTlong/aida'
If you have feedback or need assistance with the MCP directory API, please join our Discord server
