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.
npx agentmods add instructions/w8123/enterpriseagentframework/agents-mdgit clone --depth 1 https://github.com/w8123/EnterpriseAgentFrameworkWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.02764 | $0.02764 |
| Opus 5 | $0.01382 | $0.01382 |
| Sonnet 5 | $0.00553 | $0.00553 |
| Haiku 4.5 | $0.00276 | $0.00276 |
Grade A, and why
EnterpriseAgentFramework AGENTS.md scanned grade A with 0 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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 115 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ReachAI Agent Rules
本文件是给 Codex、Cursor、Claude Code、Copilot CLI 等 AI 编程工具看的项目级规则。进入本仓库后,智能体必须先读本文件,再执行代码、SQL 或文档修改。
工作原则
- 以当前仓库为事实源。先查真实代码、SQL、接口、前端页面和文档,再下结论。
- 可以修改前端、后端、SQL、文档和构建配置,但改动必须聚焦当前任务。
- 不要为了兼容旧数据而牺牲当前正确设计。本项目默认处于快速迭代阶段,schema 和种子数据可以按当前目标直接演进。
- 不要回滚、格式化或重写与任务无关的用户改动。
- Windows PowerShell 是常见执行环境;中文文件统一按 UTF-8 读写,避免 GBK/mojibake。
- 使用
rg/rg --files优先查找文件和文本。
Windows PowerShell 与原生程序 UTF-8 硬约束
-
含中文的 Python、Node、JSON、SQL 或请求体,禁止直接通过 Windows PowerShell 默认管道传给原生程序,例如
@'...中文...'@ | python -。 -
优先使用
apply_patch创建 UTF-8 临时脚本或 JSON 文件,再执行对应程序。 -
如果必须通过 PowerShell 管道传递中文,执行前必须设置:
$utf8 = [System.Text.UTF8Encoding]::new($false) $OutputEncoding = $utf8 [Console]::InputEncoding = $utf8 [Console]::OutputEncoding = $utf8 -
向数据库或 API 写入中文后,必须回读验证;如果中文变成连续
?,视为执行失败,不得继续后续验收。
项目定位
ReachAI 是面向 Java 企业系统的 AI 能力中台,不只是 Workflow Builder,也不只是扫描历史项目生成 Tool。
当前主线是:
- 业务系统通过
reachai-spring-boot2-starter、reachai-capability-sdk、@ReachCapability、@ReachParam主动注册项目、实例、能力和 SDK 图。 - 平台侧形成能力快照、字段级 diff、评审 apply/ignore,并沉淀到能力资产目录。
- Workflow Studio 使用
GraphSpec编排 Workflow(表:runtime_workflow);Agent(表:runtime_agent)通过已发布配置版本和 Workflow-as-Tool 白名单,由 Supervisor 按用户意图选择零个、一个或多个 Workflow。AI 生成、局部修改、调试、发布和回放均在 Workflow Studio 闭环。 - Runtime 通过
SupervisorRuntimeAdapter承载 AgentScope 的理解、规划、选择与有限重规划,通过AgentRuntimeAdapter/ GraphSpec executor 承载 Workflow 执行,并为未来运行时保留适配边界。 - RunOps、Trace、Tool ACL、Guard、Gateway、MCP、A2A、嵌入式对话和企业身份共同组成生产治理边界。
当前后端重塑已进入物理服务拆分后的旧结构退场阶段。目标拓扑是 reachai-control-service、reachai-runtime-service、reachai-capability-service、reachai-knowledge-service、reachai-model-service。第一阶段保持同一个 MySQL 库,不拆库;公共入口由 reachai-control-service 保持 /api/**、/embed/** 和 SDK 注册入口兼容。旧 ai-agent-service module 已从仓库主路径删除,不再作为 Maven、IDEA、本地启动或部署单元存在。任何新增代码必须遵守服务表所有权和服务间 API 边界,不允许为了快速编译跨服务直接写对方表。
关键目录
ai-admin-front/: Vue 3 + Element Plus 管理端。reachai-control-service/: 当前 Platform Control / public API BFF 主入口,保持/api/**、/embed/**和 SDK 注册公开入口。reachai-runtime-service/: 当前 Runtime Host 部署单元,承接 Agent、Workflow、GraphSpec、Trace、RunOps、调试和运行时内部 API。reachai-capability-service/: 当前 Capability Catalog 部署单元,承接 SDK 注册、能力快照、diff/review/apply、扫描目录和能力资产 API。reachai-knowledge-service/: 当前 Knowledge / Retrieval 部署单元;包括知识库、文件、chunk、RAG、业务索引、向量检索和历史扫描器实现。不要再把它描述成“技能服务”。reachai-model-service/: 当前 Model Gateway 部署单元;包括模型中心 V2(model_template+model_instance)、Chat、Embedding、Rerank。不再提供未使用的/model/openai-proxy入口。reachai-capability-sdk/: JDK8 兼容的业务能力声明 SDK 契约。reachai-spring-boot2-starter/: Spring Boot 2 业务系统接入、扫描、注册和 SDK 图同步。ai-runtime-contract/: 中台内部业务记忆引用契约。sql/initV2.sql: 当前新库 SQL 基线入口;第一阶段仍是同一个 MySQL 库,不拆库。sql/upgrade-*.sql: 仅用于当次数据库变更升级已有开发/测试库;当前新库基线不要求兼容旧数据。docs/: 当前权威知识库。docs/ai-memory/: 给 AI 编程工具看的项目记忆。
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.
- 2d ago First seen · 115 lines · 2,764 tokens per session scan A 758225948b4b
EnterpriseAgentFramework AGENTS.md is an instructions file published in the GitHub repository w8123/EnterpriseAgentFramework (618 stars, last pushed 9d ago), licensed MIT. It adds 2,764 tokens to every session, about $0.0138 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
OmniRoute CLAUDE.md
Claude Code instructions for diegosouzapw/OmniRoute, covering claude.md, worktree isolation — claude code specifics, cross-session safety — claude code specifics, superpowers / planning artifacts — path overrides and scratch / temporary files — use artifacts/, not /tmp.
OmniRoute AGENTS.md
AGENTS.md instructions for diegosouzapw/OmniRoute, covering omniroute agent guide, quick start, running tests, single test file (node.js native test runner — most tests) and vitest (mcp server, autocombo, cache).
OmniRoute copilot-instructions.md
Copilot instructions for diegosouzapw/OmniRoute, a project described as: Never stop coding. Free MIT AI gateway: one endpoint, 352 providers (150+ free), 1200+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback…
OmniRoute GEMINI.md
Gemini CLI instructions for diegosouzapw/OmniRoute, a project described as: Never stop coding. Free MIT AI gateway: one endpoint, 352 providers (150+ free), 1200+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback…
ag2 AGENTS.md
AGENTS.md instructions for ag2ai/ag2, covering ag2 development guidelines, ai-assisted contribution policy, architecture decision records (adr), code style guidelines and package structure.
agent-framework-js CLAUDE.md
Instructions for polymind-inc/agent-framework-js, covering notes for ai coding assistants, non-negotiable rules, platform and status.