OthmanAdi/planning-with-files
Persistent file-based planning for AI coding agents and long-running agentic tasks. Crash-proof markdown plans that survive context loss and /clear, plus a deterministic completion gate and multi-agent shared state on disk. Manus-style. Works with Claude Code, Codex CLI, Cursor, Kiro, OpenCode and 60+ agents via the SKILL.md standard.
Python · ★ 24,791 · Fork 2,109 · 在 GitHub 查看
在手机上浏览源码、订阅每日 AI 趋势日报与项目架构深度解读:获取 CopoHub 客户端(iOS & HarmonyOS)→
GitHub Trending 上榜记录
OthmanAdi/planning-with-files 已在 GitHub Trending 每日榜单上榜 2 天,最高排名第 20,首次上榜 2026-07-05,最近一次上榜 2026-07-06。
| 日期 | 排名 | Star | 当日增长 |
|---|---|---|---|
| 2026-07-06 | #20 | 24,791 | +66 |
| 2026-07-05 | #20 | 24,665 | +61 |
同期上榜的项目
- asgeirtj/system_prompts_leaks – Leaked system prompts of ChatGPT, Claude, Gemini and more. The hidden system prompt of every popular chatbot, in full, every version since May 2025. Also Anthropic Claude Code, OpenAI Codex, Google Antigravity, xAI Grok, Meta Muse, Moonshot Kimi, Alibaba Qwen, Z.ai GLM ZCode, MiniMax, Cursor, Perplexity, DeepSeek and 30+ others. Updated daily.
- addyosmani/agent-skills – Production-grade engineering skills for AI coding agents.
- Zackriya-Solutions/meetily – Privacy first, AI meeting assistant with 4x faster Parakeet/Whisper live transcription, speaker diarization, and Ollama summarization built on Rust. 100% local processing. no cloud required. Meetily (Meetly Ai - https://meetily.ai) is the #1 Self-hosted, Open-source Ai meeting note taker for macOS & Windows. Understand How to write meeting minutes
- ruvnet/RuView – π RuView turns commodity WiFi signals into real-time spatial intelligence, vital sign monitoring, and presence detection — all without a single pixel of video.
- Leonxlnx/taste-skill – Agent skills that give coding agents taste: art direction, reference images and a browser check instead of generic AI pages
- alirezarezvani/claude-skills – 380 Claude Code skills & agent skills & plugins (30+ Agents, 70+ custom commands, 380+ skills, customizable references, scripts)for Claude Code, Codex, Gemini CLI, Cursor, and 8 more coding agents — engineering, marketing, product, compliance, C-level advisory, research, business operations, commercial & finance, and your daily productivity skills.
planning-with-files 项目深度剖析与商业评估报告
planning-with-files 项目深度剖析与商业评估报告
1. 核心定位与商业落地价值
根本痛点:AI Agent 的“记忆失忆症”
该项目精准切入 AI 编码 Agent 最核心的工程瓶颈:上下文窗口的易失性与长期任务执行的矛盾。当前主流 LLM Agent 架构中,TodoWrite 等工具调用在上下文重置后即消失,导致: 目标漂移:50+ 工具调用后,Agent 遗忘原始目标 错误重复:失败日志不可持久化,无法进行根因分析 上下文膨胀:将中间状态全部塞入窗口,加速上下文退化
该项目将“文件系统即磁盘”的模式引入 Agent 工作流,本质上是将 Agent 的“工作内存”从易失的 LLM 上下文窗口迁移至持久化文件系统,这是 Manus 在 8 个月内做到 $100M+ 营收的核心工程秘密。
企业级商业落地场景
| 场景 | 价值量化 | 关键能力 | |------|---------|---------| | CI/CD 管道中的长期自动化测试 | 减少 60%+ 人工介入 | 崩溃恢复、多阶段完成门控 | | 多 Agent 协作代码审查 | 并行任务隔离,避免状态冲突 | 隔离计划目录、SHA 校验 | | 金融/合规场景的审计追踪 | 满足 SOC2 对 Agent 行为的可追溯性要求 | 不可变计划、JSONL 运行分类账 | | DevOps 事故响应 | 从“人工驱动”转为“Agent 自主编排” | 自治模式、阶段依赖图 |
竞品差异化与护城河
在 README 中展示上榜徽章
[](https://github.fq6825.top/repo/OthmanAdi/planning-with-files)