agentmods.dev

ccteam AGENTS.md

Repository instructions for coding agents working on ccteam, a service that connects chat systems with multiple coding-agent sessions.

In plain words
What is it for?
Use it when changing ccteam, especially when tracing sessions, agent communication, command-line behavior, routing, or local project documents.
Why use it?
It tells agents where to find current requirements, how to start work, and which architecture rules must not be broken.

Instructions file for CodexOpenCode

377 2d ago D scan 7,450 tokens original MIT
View the source file firstintent/ccteam
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/firstintent/ccteam/agents-md
Clone the repo
git clone --depth 1 https://github.com/firstintent/ccteam

Made for: Codex, OpenCode.

Per session 7,450 This file is loaded in full into every session.
When invoked 7,450 The same file — it is already loaded in full.
Security scan D 3 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.07450 $0.07450
Opus 5 $0.03725 $0.03725
Sonnet 5 $0.01490 $0.01490
Haiku 4.5 $0.00745 $0.00745

Measured 2d ago against content hash 90f5afd4b9a1, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade D, and why

ccteam AGENTS.md scanned grade D with 3 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Downloads and executes remote codehighSupply chain

curl | sh runs whatever the server returns today, which is not necessarily what it returned when this was reviewed.

- **安装面 = 一个二进制引擎、两个安装/托管面**:CLI(`curl install.sh | sh` → `ccteam config` 注册五 vendor 全局 MCP;pi 例外走受管会话 bridge)与 DSH 插件 `@ccteam/ccteam-ui`(npm,自带引擎平台包 `@ccteam/engine-<os>-<cpu>`,`apply()` 自动装/自启/attach);二者共用同一 `$CCTEAM_

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

| **vendor 配置足迹 = 只写自家 MCP 注册** | 对 vendor 全局配置(`~/.claude.json`、`~/.codex/config.toml`、grok/opencode/kimi 对称面)的唯一写入 = 幂等注册/修复自己的 MCP server 条目;项目侧只写 `.claude/settings.local.json` 自己的段(绝不碰用户 `.claude/settings.json`);pi 不

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

- **安装面 = 一个二进制引擎、两个安装/托管面**:CLI(`curl install.sh | sh` → `ccteam config` 注册五 vendor 全局 MCP;pi 例外走受管会话 bridge)与 DSH 插件 `@ccteam/ccteam-ui`(npm,自带引擎平台包 `@ccteam/engine-<os>-<cpu>`,`apply()` 自动装/自启/attach);二者共用同一 `$CCTEAM_
AGENTS.md · 117 lines

How it starts

The opening of the file, as written. The whole thing — 117 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AGENTS.md — ccteam 实现导引

权威文件 = AGENTS.md;CLAUDE.md 是指向它的软链。面向下一次接手 ccteam 的 agent session(Claude Code 读 CLAUDE.md,Codex 读本文),起手必读。 代码是唯一事实源:协议 / CLI / JSON / event / 路由 / 参数一律以代码与工具自描述为准;本文只留定向图 + 红线 + 纪律 + 实战坑,不复述代码能回答的事。 需求与发版由 docs-local/(gitignored 本机文档区)管理(§一):版本 PRD 自上而下 + 本地 issue 自下而上;仓库内不再维护任何治理状态文件(.loop/ 已于 2026-08-30 退役,勿复活)。 冷启动:本文 → git log + Cargo.toml 版本 → docs-local/versions/<当前版本>/prd.md + docs-local/issues/(如有)→ 代码按需读,不做全仓扫描。

〇、架构总览(定向图;红线唯一清单 = §三)

ccteam = 多 harness agent 团队的桥接与治理层:常驻 daemon(IM gateway + web + MCP)把 IM/web chat 路由到按需 spawn/resume 的 agent session;任意 session 经 8 个 MCP 工具委派任意其他 session(A2A)。铁律:只做单 harness 做不到的(跨 vendor 身份/路由/账本/观测 + 跨机执行),永不做厂商能力。

  • 核心模型 chat ⇄ project ⇄ session:session 是一等实体(持久 sid s<N>,单调、扛重启、不复用);role / harness(claude|codex|grok|opencode|kimi|pi|dsh,AgentVendor 可扩展)/ provider(model)/ protocol 都是 session 属性,同一 role 可并存多 session;host 是 project 属性,session 继承;roleless(空 role,裸 vendor 自读项目 CLAUDE.md)合法且是默认。
  • 协议轴:Claude 默认 stream-json(长驻子进程 + 双向 NDJSON,无 PTY/hook);grok/opencode/kimi/dsh 走 ACP;pi 走 --mode rpc(LocalOnly);terminal(tmux/rmux)维护期、规划淘汰(§三)。所有 adapter 归一 CanonicalEvent,gateway spawn_event_pump 单点消费。
  • 数据面:业务事件 = progress.jsonl(schema 权威 harness/progress_bridge,core 只 re-export);对话原文 = <project>/.ccteam/chat/<sid>/turns.jsonl(按 sid;live daemon 唯一 writer = gateway);成本/委派树全入账本。
  • 接口面:8 个 MCP 工具 mcp__ccteam__{status(+裸名发现别名 grok_claude_codex_kimi),chat_send_file,session_*×5} + POST /mcp streamable HTTP + REST /api/v1(OpenAPI = /api/docs)+ IM 斜杠命令 + chat-shell web(per-session Chat|终端);清单/参数/语义以代码与工具自描述为准。
  • 内容面:引擎零内置 persona/skill/提示词;role = 项目 .claude/agents/<role>.md(vendor 原生 --agent 自读);skill = 全局库 ~/.ccteam/skills(会话显式 attach)+ 项目 .agents/skills(.claude/skills 软链);一切内容从 ccteam-hub 装(sha256 校验、never-execute)或用户自建。
  • 执行面:daemon 不 tick、无 orchestrator 循环,只响应消息/排程;会话 = resume-by-sid + 空闲释放 + 容量挤停;编排智能 100% 用户空间(ccteam-flow/workflow.yaml 占位 deferred)。
  • 安装面 = 一个二进制引擎、两个安装/托管面:CLI(curl install.sh | sh → ccteam config 注册五 vendor 全局 MCP;pi 例外走受管会话 bridge)与 DSH 插件 @ccteam/ccteam-ui(npm,自带引擎平台包 @ccteam/engine-<os>-<cpu>,apply() 自动装/自启/attach);二者共用同一 $CCTEAM_HOME 与同一个 daemon(谁先起谁赢、其余 attach;插件释放不停 daemon;/health 身份字段 + run/daemon-endpoint.json pid 门)。

Read the full file on GitHub · 117 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 117 lines · 7,450 tokens per session scan D 90f5afd4b9a1

Subscribe to this mod's changes

ccteam AGENTS.md is an instructions file published in the GitHub repository firstintent/ccteam (377 stars, last pushed 2d ago), licensed MIT. It adds 7,450 tokens to every session, about $0.0372 per session on Opus 5. A static security scan graded it D with 3 findings (downloads and executes remote code, reads agent configuration directories, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Writing and docsagentclaude-codecodexdeepseek-harnessgrokkimi-code
Related

Other instructions, from other repositories

dsh-worktable AGENTS.md

Instructions for Aisland-SJL/dsh-worktable, covering dsh-worktable 项目规则, 协作方式（用户定案，最高优先级）, 边界, 构建与验证 and 领域约定（会话中必须遵守）.

Aisland-SJL/dsh-worktable · 4,863 tokens

cocode AGENTS.md

Instructions for cocode-agency/cocode, covering agents.md, 仓库布局, 仓库边界, cocode-gui/ and cocode-tui/.

cocode-agency/cocode · 1,463 tokens

neeko AGENTS.md

Instructions for tincopper/neeko, covering neeko — repository guidelines, 项目概览, 目录结构, 前端（feature-based 架构） and 后端（domain-driven 模块化架构）.

tincopper/neeko · 7,575 tokens

grok-delegate AGENTS.md

Instructions for arthurkatcher/grok-delegate, covering agents.md — grok-delegate, what this repo is, layout (do not invent a second one), non-negotiables and how to work here.

arthurkatcher/grok-delegate · 908 tokens

neeko CLAUDE.md

Instructions for tincopper/neeko, covering neeko — claude.md, 项目概览, 常用命令速查, 键盘快捷键 and 预置 agent & ide.

tincopper/neeko · 790 tokens

clique CLAUDE.md

Claude Code instructions for thejdubb02/clique, covering clique — standing rules, motion, looking at it, the api is the whole surface and releasing.

thejdubb02/clique · 1,982 tokens