Basic
$0/month
For individual developers proving the integration.
- 2,000 security decisions
- 1 agent
- 2 scoped API keys
- Approvals and incidents
- Tamper-evident audit
Simple monthly plans
Start free. Scale by secured decisions.
Every plan includes the same core authorization, approvals, incidents and audit capabilities. Limits change with your workload.
Plus
$9/month
For small production agents and pilots.
- 50,000 security decisions
- 3 agents
- 5 scoped API keys
- Approvals and incidents
- Tamper-evident audit
Premium
$29/month
For software teams operating multiple agents.
- 500,000 security decisions
- 10 agents
- 20 scoped API keys
- Approvals and incidents
- Tamper-evident audit
Pro
$79/month
For high-volume agent services.
- 2,000,000 security decisions
- 30 agents
- 50 scoped API keys
- Approvals and incidents
- Tamper-evident audit
A decision is one unique authenticated authorization operation. Idempotent retries are not counted twice. Automatic overage billing is disabled.
Questions before you connect
Does a denied operation count?
Yes. ALLOW, DENY and REQUIRE_APPROVAL are valid security decisions. Invalid credentials, health checks and rate-limit rejections do not count.
Does AgentSecurity charge for LLM tokens?
No. The deterministic authorization path does not invoke an LLM. Your model provider charges remain separate.
What happens at the monthly limit?
New decisions fail closed with a quota response and reset date. We do not create surprise overage charges.
Are the integrations different by plan?
No. Supported Python, LangChain, LangGraph, CrewAI and OpenAI Agents adapters use the same enforcement features on every plan.