sofagent AGENTS.md

An AGENTS.md file that connects sofagent to Codex and defines a four-level loading order for skills, user rules, reflection notes, and knowledge files.

In plain words
What is it for?
Use it to configure sofagent sessions, load project guidance, connect its MCP server, restrict tool roles, and run repository audits.
Why use it?
It tells the agent where its governing instructions come from and requires audits of work across platforms.

Instructions file for CodexOpenCode

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/kongfangxun/sofagent/agents-md
Clone the repo
git clone --depth 1 https://github.com/KongFangXun/sofagent

Made for: Codex, OpenCode.

Per session 562 This file is loaded in full into every session.
When invoked 562 The same file — it is already loaded in full.
Security scan B 1 finding. 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 $0.00562 $0.00562
Opus 5 $0.00281 $0.00281
Sonnet 5 $0.00112 $0.00112
Haiku 4.5 $0.00056 $0.00056

Measured 2d ago against content hash cbd1857df88d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

sofagent AGENTS.md scanned grade B with 1 finding 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.

Reads agent configuration directoriesmediumAgent snooping

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

# ~/.codex/config.toml
AGENTS.md · 39 lines

What it actually says

AGENTS.md · sofagent Codex 适配(薄挂载)

本文件是 Codex CLI 的顶层指令挂载点(AGENTS.md 约定),不含规范内容——四层加载链的规范资产唯一来源是 sofagent 的 SKILL/ 目录(单一真相源)。规则更新只发生在 SKILL/,本文件永不复制它们。

区分:FORGE/SKILL/ 不是规范资产——它仅存放 FORGE 内部循环(fresh-eyes-loop、release-gate-loop)的技能定义,与根 SKILL/ 规范目录同名但职责无关。

会话开始时按序加载(四层加载链)

一、读取已安装的 SKILL.md 作为系统约束(L1 硬约束):

~/.sofagent/skills/sofagent/SKILL.md

(若用 bash install.sh --platform codex 安装,fde.md 已写入 ~/.codex/fde.md。)

二、读取用户工作规则(L2):~/.codex/fde.md(或 ~/.sofagent/skills/sofagent/fde.md

三、读取反思区(L3):~/.sofagent/skills/sofagent/think.md

四、知识库(L4):~/.sofagent/skills/sofagent/knowledge/ 目录,按任务关键词检索加载。

model_instructions_file(可选)

Codex 的 model_instructions_file 配置可直接指向 SKILL.md,让约束进系统层:

# ~/.codex/config.toml
model_instructions_file = "~/.sofagent/skills/sofagent/SKILL.md"

审计强制(平台无关)

约束是建议性的,审计是强制性的——提交前审计走 git hooksofagent-audit --install-hook),与宿主平台无关:24 条 git diff 规则 + HMAC 链审计在 Codex 下同样生效。

连接 MCP Server

76 个 tool(run_audit / worklog_query / snapshot_restore / fde_interview 等)经 MCP 协议面暴露——适配器只依赖 @public API 子集,@internal 破坏性变更不影响本挂载。默认全量暴露,SOFAGENT_MCP_ROLES 显式收窄专职面。

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 · 39 lines · 562 tokens per session scan B cbd1857df88d

Subscribe to this mod's changes

sofagent AGENTS.md is an instructions file published in the GitHub repository KongFangXun/sofagent (41 stars, last pushed 4d ago), licensed MIT. It adds 562 tokens to every session, about $0.0028 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.