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/leenkcool/humanasllm/claude-mdgit clone --depth 1 https://github.com/leenkcool/humanasllmWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/leenkcool/humanasllm/claude-md)<a href="https://agentmods.dev/instructions/leenkcool/humanasllm/claude-md"><img src="https://agentmods.dev/badge/instructions/leenkcool/humanasllm/claude-md.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.02052 | $0.02052 |
| Opus 5 | $0.01026 | $0.01026 |
| Sonnet 5 | $0.00410 | $0.00410 |
| Haiku 4.5 | $0.00205 | $0.00205 |
Grade A, and why
humanasllm CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 114 lines — stays where its author put it; the contents beside it link to each section on GitHub.
P390 · Human as Agent（人即智能体）
项目定位
让工程师作为「人肉大模型」接入现有多模型 Agent 调度框架。对外完全兼容 OpenAI 标准接口（/v1/chat/completions），上游调度池新增一条 human-llm 模型路由即可接入，零代码改动。涉密/私有逻辑任务经此路由派发给人工工程师，完成后按大模型格式返回，AI 工作链路不中断。
技术栈
- 后端：Node.js + Express + Socket.IO
- 数据库：PostgreSQL 5433（库名
p390，用户 postgres） - 认证：JWT（jsonwebtoken + bcryptjs）
- 前端：Vanilla HTML/JS/CSS（IIFE 模块 + SVG 图标 + 4 套主题）
目录结构
server.js # 入口：Socket.IO 认证 + 安全头 + 路由 + 超时扫描
db/
index.js # PG 连接 + 建表（users/tasks/task_logs/request_logs）
adapters/pg.js # pg 适配器（?→$n、INSERT RETURNING id）
dialect.js # 方言工具
middleware/
auth.js # authenticate / signToken / requireRole
security.js # 基础安全头（无 CSP/HSTS）+ CORS + 限流
routes/
v1.js # OpenAI 兼容：/v1/models + /v1/chat/completions + /v1/tasks/:id 回查
auth.js # 登录/me/改密
users.js # 工程师账户管理
tasks.js # 任务列表/详情/状态流转
workbench.js # 统计/治理概览/我的任务/待接单/未完成聚合
approvals.js # 审批列表 + 批准/驳回
projects.js # 项目管理 + 申请建项目
logs.js # 请求出入日志 + 任务审计
rules.js # 分级规则管理（治理配置后台）
tenants.js # 租户管理（多租户）
gateway.js # 网关配置 + 多工具 SKILL/AGENT 生成/微调/安装
audit.js # 合规报告 + 质量数据资产导出
prd.js # PRD 需求沉淀（git 本地提交）
index.js # 路由总挂载
services/
stateMachine.js # 状态机单例（任务/审批转换表 + 校验）
categoryEngine.js # 分级策略引擎（规则白名单锁死 > 上游显式 > 默认）
queueService.js # 任务状态机流转 + AI 降级 + 质量校验 + 30s 超时扫描
aiShift.js # 智能漂移（general 简单任务 AI 承接，confidential/ops 锁死）
approvalService.js # 审批状态机 + 挂起等待 + 24h 超时提醒
aiRelay.js # DeepSeek 中继（一次 + SSE 透传）
openaiEncoder.js # OpenAI 请求解析 + 响应/SSE chunk 封装
projectService.js # 项目 CRUD + 审批批准回调
notifier.js / mailer.js # 通知（邮件 + Webhook），可降级
waiters.js # 等待者单例（审批挂起等待）
i18n.js / csv.js # 消息翻译 / CSV 导出
websocket.js # Socket.IO 推送（task:new/update/timeout）
public/ # 落地页(landing/) + 登录页 + 工作台(workbench.html + utils/api/ws/ui/app/i18n.js)
scripts/seed.js # 种子账户
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.
- 6d ago First seen · 114 lines · 2,052 tokens per session scan A 82e09ec85646
humanasllm CLAUDE.md is an instructions file published in the GitHub repository leenkcool/humanasllm (28 stars, last pushed 20d ago), licensed MIT. It adds 2,052 tokens to every session, about $0.0103 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
claude-tap AGENTS.md
Instructions for liaohch3/claude-tap, covering maintainer automation notes, agents 索引, documentation boundary, review guidelines and pre-commit hook.
opencodex AGENTS.md
AGENTS.md instructions for lidge-jun/opencodex, covering agents.md, what this project is, repository layout, optional subsystems stay off the core path and the devlog directory.
goai AGENTS.md
AGENTS.md instructions for zendev-sh/goai, covering agents.md - goai, commands, architecture, key rules and adding providers.
reddit-research-mcp CLAUDE.md
Claude Code instructions for dialog-tools/reddit-research-mcp, covering reddit research mcp server - development guide, project overview, key files, releasing new versions and automated release process.
Agent-Span AGENTS.md
AGENTS.md instructions for oxbshw/Agent-Span, covering agents.md, build & test, crate layout, adding a channel and adding an mcp tool.
gitlab-mcp-server mcp-best-practices.instructions.md
MCP protocol-level best practices for tool design, annotations, response formats, pagination, and security. Applies to all Go MCP server code.