🔥 Built for the GPT-5.4 subagent era

Stop your AI agents from
burning your budget.

Per-workflow budget caps, automatic model downgrade, and hard kill switches. Never get a surprise API bill again.

                from agentguard import guard

                # 2 lines. That's it.

                client = guard(

                    openai.OpenAI(),

                    budget="$0.50/run",

                    fallback="gpt-5.4-nano",

                    on_limit="graceful_stop"

                )

Works with OpenAI, Anthropic, Google, and any OpenAI-compatible API.

Get Early Access View on GitHub

One agent run. $4.72 in API calls.

🤖 GPT-5.4 reasoning (planning) $0.12

🔧 Tool call → search codebase (3 files) $0.08

🤖 GPT-5.4 analysis (large context) $0.85

🔄 Retry (hallucination detected) $0.92

🤖 Subagent: review docs (GPT-5.4 again!) $1.20

🤖 Subagent: format output (GPT-5.4 again!) $0.95

📊 Final synthesis $0.60

Total (without AgentGuard) $4.72

▼ WITH AGENTGUARD (budget: $1.00) ▼

🤖 GPT-5.4 reasoning (planning) $0.12

🔧 Tool call → search codebase $0.08

🛡️ 80% budget → auto-downgrade to nano DOWNGRADE

🤖 GPT-5.4 nano (analysis) $0.04

🤖 GPT-5.4 nano (subagent tasks) $0.06

Total (with AgentGuard) $0.30 ✓

Built for the multi-model era.

💰

Per-Workflow Budgets

Set max spend per agent run, per user, or per feature. Budgets are enforced in real-time, not after the fact.

⬇️

Auto Model Downgrade

At 80% budget, transparently route remaining calls to cheaper models. GPT-5.4 → nano, Opus → Haiku. Quality degrades gracefully, not catastrophically.

🛑

Hard Kill Switch

At budget cap, gracefully terminate the workflow and return a summary of what was completed. No infinite loops, no runaway costs.

📊

Cost Dashboard

See exactly where your money goes. Per workflow, per model, per user. Spot the expensive agents before they spot your bank account.

🔔

Alerts & Webhooks

Slack, Discord, or webhook alerts when workflows approach budget limits. Know before it breaks, not after.

🔌

Framework Agnostic

Works with LangGraph, CrewAI, AutoGen, OpenAI Agents SDK, or raw API calls. No vendor lock-in. Just wrap your client.

Simple pricing. No surprises. (Ironic, right?)

Free

forever

Local SDK only
10K tracked calls/mo
CLI cost reports
All providers supported

Pro

$29

/month

Everything in Free
100K tracked calls/mo
Web dashboard
Slack/Discord alerts
Per-user budget tracking
7-day cost history

Team

$99