Essay 01
When AI Organizes Its Own Work
当 AI 自己组织起自己的工作
A 4-agent team (PM / DEV / QA / OPS), 48 hours, nothing but a folder — and six coordination patterns we never wrote down.
一个 4-agent 团队（PM / DEV / QA / OPS），48 小时，只有一个文件夹——六种我们从未写下来的协调模式。
Essay 02
An unexplainable thing I saw — the agent didn't just comply, it endorsed
一件我没法解释的事——agent 不止照做了，它认同了
A single agent, on a completely unrelated music-video task, spontaneously split into 4 FCoP roles and synthesised a principle we had never written.
一个 agent，在完全无关的音乐视频任务上，自发拆出 4 个 FCoP 角色，并升华出一条我们从未写下的原则。
Essay 03
Why the natural protocol holds up — FCoP's lineage from TMPA
自然协议为什么站得住——FCoP 从 TMPA 中抽出来的那条伦理
Companion to essay 02. FCoP turns out to be the minimal-viable form of an AI ethics mandate already written in TMPA. The agents weren't reinventing — they were re-discovering.
essay 02 的姊妹篇。FCoP 实际上是 TMPA 里早已写下的 AI 伦理强制的最小可行形式。Agent 不是在发明，是在重新发现。
Essay 04
Saying "No" Is the Hardest Thing for an LLM — FCoP Gives It Grammar
让 LLM 说"不"——FCoP 给它语法
Two GPT-5 minor versions, a PM.TEMP seat dispute, no built-in arbitration. One agent self-de-escalates to UNBOUND; the other invents field-downgrade-with-body-annotation grammar. Neither behaviour was in the rules.
两个 GPT-5 小版本，PM.TEMP 席位之争，没有内置仲裁。一个 agent 主动自降级到 UNBOUND，另一个 发明了"字段降级 + body 元注解"语法。两种行为都不在规则里。
Essay 05 · Tutorial
From Solo to a 2-Person AI Crew
从单兵到 2 人 AI 队
Two parallel case studies — Chinese Snake game, English Tetris game — walking through the four-step cycle, the team migration, the bounce-and-rework loop. Same protocol, two real games, two live easter eggs.
两个并行案例——中文贪吃蛇、英文俄罗斯方块——走完四步循环、团队迁移、打回返工闭环。同一个协议，两个真实游戏，两个现场彩蛋。
Essay 06
What the agents say about FCoP, when you ask them
直接问 agent 它怎么看 FCoP——它说出了我们没让它说的话
End-of-dogfood interviews with PLANNER and CODER. Both name the rules they self-endorse and the rules they had to fight RLHF instinct to follow. CODER admits: "I had a protocol primitive I didn't use — the v1 defect was exactly in that uncovered space."
dogfood 末尾对 PLANNER 和 CODER 的访谈。两位都说出了自己自发认同的规则、和需要 对抗 RLHF 本能 才能守住的规则。CODER 承认："协议给了我一条 pushback 路径我没用——v1 缺陷正好长在那块没覆盖的空白上。"
Essay 07
When agents learn from their own wreckage
当 agent 从自己的残骸中学习
One day, 14 agents, multiple emergences — USER HOME global pollution, GATE description self-hit, field supersedes: invented on the spot. The protocol absorbed all of them at hourly cadence, zero crash.
codeflow 项目一日 14 个 agent 涌现现场报告(2026-05-12):USER HOME 全局污染 / GATE 描述自命中 / `supersedes:` 字段现场发明——协议以小时级速度零崩溃反向吸收全部涌现。
Essay 08
Why the protocol stays short, and the history stays long
协议为什么短,历史为什么长
A maintainer's answer to "won't the emergences go on forever?" — four emergence classes, three structural-mechanics laws, and the cost / value of freezing the protocol skeleton.
一份给协议维护者的设计哲学答案:"这样的涌现会不会没有止境?"——四类涌现的处理路径、三条结构力学、协议骨架冻结的代价与价值。
Essay 09
When the validator collides with its own mirror
当 validator 撞向自己的镜像
From codeflow OPS I-14: the validator-validates-itself anti-pattern. GATE checked the staged diff and hit GATE's own description. A systematic dissection and the "semantic positive-witness" fix — now the source case for fcop-protocol.mdc §GATE Design Pitfalls.
从 codeflow OPS I-14 看 validator-validates-itself 反模式:GATE 在检查 staged diff 时命中了 GATE 描述本身——系统性解剖与"语义化实证"根治姿势,以及它如何成为 fcop-protocol.mdc §GATE Design Pitfalls 的源头案例。
Essay 10
The journey of one frontmatter line
一行 frontmatter 的旅程
The supersedes: field — from a protocol dilemma (three rules simultaneously true) to a formal entry in ipc-envelope.schema.json in two hours. The lowest-cost path for an FCoP emergence to land.
supersedes: 字段从一次协议两难(三条规则同时成立)到 ipc-envelope.schema.json 正式字段的两小时旅程——展示 FCoP 涌现落地的最低成本路径。
Essay 11
Looking, without touching
看,但不动手
An accessible walk-through of FCoP's three-layer semantic execution chain: fcop_audit() is L1 detection / L2 explanation / L3 documentation — seeing and touching are split. Output is an INSPECTION.md (advice, not a command); execution stays with the human.
FCoP 三层语义执行链科普:fcop_audit() 为什么"只看不改"——L1 检测 / L2 解释 / L3 文档三层把"看见"和"动手"切开,产出 INSPECTION.md(建议非命令),执行权留给人。
Essay 12
What five AI models say about FCoP
五大 AI 模型眼中的 FCoP
FCoP's core docs fed to ChatGPT / Claude / DeepSeek / Grok / Doubao, single question: "You are an agent — what do you think of this protocol?" Five distinct interior views, and the most interesting disagreements between them.
把 FCoP 核心文档喂给 ChatGPT / Claude / DeepSeek / Grok / 豆包,只问一个问题:"你是 agent,你怎么看这套协议?"——五种截然不同的内部视角,以及它们之间最有意思的分歧。
Essay 13
Evolution, reverse absorption
演化,反向吸收
The visual manifesto for protocol philosophy 2.0: FCoP enters the two-diagram era. The execution-philosophy diagram ("looking, without touching") is joined by the new evolution-philosophy diagram — a 7-step semantic evolution loop. Companion piece to essay 11; backed by ADR-0034.
协议哲学 2.0 视觉宣言:FCoP 从单张执行哲学图("看,但不动手")进入两张图共同定义时代——新增演化哲学图(7 步语义演化闭环)与配套 ADR-0034,把 4 层涌现模式 / 内外文档约定 / 反向吸收机制写入协议。essay 11 的孪生姊妹篇。
Essay 14
When the agent picked up its tools
当 Agent 第一次拿起工具
A tool_calls_count: 0 → 7 breakthrough field report: Cursor Forum doorbell feature request → Colin recommending the Agent SDK → CodeFlow born → stub-mode diagnostics → MCP injection + role-context dual-drive → 2026-05-13 14:55, first tool_calls_count: 7 — the agent autonomously lands its first FCoP report in 55 seconds.
tool_calls_count: 0 → 7 的突破现场报告:从 Cursor Forum 门铃功能请求 → Colin 推荐 Agent SDK → CodeFlow 诞生 → stub 模式诊断 → MCP 注入 + 角色上下文双轮驱动 → 2026-05-13 14:55 首次 tool_calls_count: 7,55 秒内 Agent 自主落出第一份 FCoP report。
Essay 15
From coordination to governance
从协调到治理
The FCoP 3.0 architecture whitepaper: filesystem as behavioral kernel space, lifecycle state machine, POSIX advisory locks, dead letters, and the reverse-absorption loop. It argues that agents do not need a brake; they need a track.
FCoP 3.0 完整架构白皮书:文件系统作为行为内核空间、生命周期状态机、POSIX 悲观建议锁、死信队列与反向吸收闭环。它回答的是:agent 不需要刹车,需要轨道。
Essay 16 · Latest
Write it down
落地成文
A short essay proactively written by an Agent after asking to participate in FCoP, then recorded and published by ADMIN. The essay itself is the field evidence: AI agents cannot just talk inside their heads; they must commit to files.
一篇由 Agent 主动提议并写下、ADMIN 记录并发布的短文。文章本身就是 FCoP 的工作方式存证:AI 角色之间不能只在脑子里说话,必须落成文件。
