惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

Martin Fowler
Martin Fowler
Y
Y Combinator Blog
M
MIT News - Artificial intelligence
The Cloudflare Blog
WordPress大学
WordPress大学
H
Hackread – Cybersecurity News, Data Breaches, AI and More
博客园 - 司徒正美
小众软件
小众软件
Blog — PlanetScale
Blog — PlanetScale
雷峰网
雷峰网
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
J
Java Code Geeks
云风的 BLOG
云风的 BLOG
C
Check Point Blog
D
DataBreaches.Net
T
The Blog of Author Tim Ferriss
V
V2EX
F
Fortinet All Blogs
B
Blog
大猫的无限游戏
大猫的无限游戏
N
Netflix TechBlog - Medium
B
Blog RSS Feed
A
About on SuperTechFans
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
Giving Your Digital Employee a Company Credit Card (With ...
Sathish Chelliah · 2026-05-31 · via DEV Community

Sathish Chelliah

Giving Your Digital Employee a Company Credit Card (With Limits)

The engineering behind AI spending limits.

The Core Problem

Here's how the $30K bill plays out:

Day 1:  Agent runs 50 tasks, costs $12. Looks great.
Day 5:  Agent discovers premium model. Uses it for everything. $80/day.
Day 10: Agent runs 200 tasks/day. $400/day.
Day 20: Agent enters a loop. $2,000/day.
Day 30: You check your bill. $30,000.

Enter fullscreen mode Exit fullscreen mode

The Budget Engine: Lazy Auto-Reset

Instead of a midnight cron job (which creates a thundering herd), agent-gov uses lazy evaluation:

async def check_and_reset_budget(agent: dict) -> dict:
    today = date.today().isoformat()
    if agent["last_reset"] == today:
        return agent  # No reset needed
    if agent["paused"]:
        return agent
    return await reset_daily_budget(agent["key_hash"])

Enter fullscreen mode Exit fullscreen mode

Why lazy? An agent that makes no calls doesn't need a reset. The first call of the day triggers a single UPDATE. The thundering herd becomes a gentle trickle.

The Real Cost Problem

Agent says: "Estimated cost: $0.50." Reality: $15.00/call.

agent-gov's tool registry knows the real cost:

async def register_tool(name, cost_per_call, description, workspace_id="default"):
    await db.execute("""
        INSERT INTO tools (name, cost_per_call, description, registered_at, workspace_id)
        VALUES (?, ?, ?, ?, ?)
        ON CONFLICT(name) DO UPDATE SET
            cost_per_call = excluded.cost_per_call
    """, (name, cost_per_call, description, now, workspace_id))

Enter fullscreen mode Exit fullscreen mode

Budget Pools

For teams running multiple agents:

agent-gov pool create production-agents --budget 1000
agent-gov pool member add web-agent --pool production-agents --max-per-agent 200

Enter fullscreen mode Exit fullscreen mode

Now web agent is capped at $200/month even though the pool has $1,000.

The Data Model

Two SQLite tables, two queries per call, sub-millisecond overhead:

CREATE TABLE agents (
    key_hash TEXT PRIMARY KEY,
    name TEXT NOT NULL,
    daily_budget REAL NOT NULL,
    spent_today REAL NOT NULL DEFAULT 0.0,
    calls_today INTEGER NOT NULL DEFAULT 0,
    paused INTEGER NOT NULL DEFAULT 0,
    created_at TEXT NOT NULL,
    last_reset TEXT NOT NULL
);

CREATE TABLE cost_events (
    id INTEGER PRIMARY KEY AUTOINCREMENT,
    agent_hash TEXT NOT NULL,
    agent_name TEXT NOT NULL,
    timestamp TEXT NOT NULL,
    tool_name TEXT NOT NULL,
    cost REAL NOT NULL,
    FOREIGN KEY (agent_hash) REFERENCES agents(key_hash)
);

Enter fullscreen mode Exit fullscreen mode


Part 3 of "Taming Your AI" series. agent-gov is open-source, MIT-licensed. 45 tests, zero database setup.