Here’s the paradox: the same bot defense systems protecting e-commerce merchants from scrapers, credential stuffers, and fraud bots are also blocking legitimate AI agents trying to buy things on behalf of real humans. Every automated request looks the same at the edge. And merchants are leaving money on the table because they can’t tell the difference.
F5 and Skyfire think the solution is an identity layer for AI agents — a verifiable credential that says “this agent is legitimate, it represents a real customer, and it can pay.”
Their partnership, announced March 18, integrates Skyfire’s Know Your Agent (KYA) protocol into F5’s Distributed Cloud Bot Defense. The result: merchants can selectively allow verified agent traffic through their existing security infrastructure without re-platforming anything.
Know Your Agent: Identity Tokens for Bots That Buy
KYA works like an agent passport. Each verified agent carries a JSON Web Token containing:
- Agent identity — what it is, who built it, what it’s authorized to do
- Principal identity — the human or enterprise behind the request
- Payment credentials — Skyfire’s tokenized payment rails, enabling the agent to complete checkout flows
The tokens use standard JWT/OAuth2/JWKS infrastructure. F5’s bot defense interprets them at the edge in real time, making enforcement decisions before traffic hits the merchant’s application layer.
This is a fundamentally different approach than blocking-by-default. Instead of treating all automated traffic as hostile, merchants define policies for verified agents — allow browsing, allow purchasing up to a budget, require human confirmation for high-value transactions.
Why This Matters Now
The timing tracks with a broader shift. AI agents are becoming autonomous consumers of web services — researching products, comparing prices, completing purchases. Agentic commerce is already happening, but the infrastructure assumes human users at the other end of every HTTP request.
The numbers make the case:
- Merchants blocking all automated traffic are blocking legitimate agent-driven revenue
- Bot defense vendors reporting that AI agent traffic is growing faster than malicious bot traffic
- E-commerce platforms seeing agent checkout attempts that can’t complete because they fail bot detection
F5’s John Maddison frames it directly: “Merchants need the ability to distinguish between a malicious agent or bot and a verified AI agent acting on behalf of a real customer.”
Agent Identity as Infrastructure
KYA joins a growing stack of agent identity solutions:
| Product | Scope | Approach |
|---|---|---|
| Okta for AI Agents | Enterprise internal | Shadow agent discovery + kill switch |
| SailPoint × AWS | Enterprise governance | Identity lifecycle on Bedrock |
| Deutsche Telekom Agent Ready | Telco-scale | Digital identities via network APIs |
| Google digital passports | Cross-enterprise | Zero trust per-action verification |
| F5 × Skyfire KYA | Internet commerce | Agent identity at the CDN edge |
The pattern: agent identity is stratifying. Internal enterprise identity (Okta, SailPoint), cross-enterprise identity (Google), telco-scale identity (Deutsche Telekom), and now internet commerce identity (F5 × Skyfire). Each layer handles a different trust boundary.
For OpenClaw users, this matters because your agents interact with the public internet. An OpenClaw agent that browses, researches, and purchases will increasingly need verifiable identity to avoid being blocked by bot defense systems. KYA tokens could become the “user-agent string” of the agentic web — a standard way to declare intent and identity at the network edge.
The Monetization Flip
The subtlest part of this partnership isn’t the security — it’s the economics. Bot defense has always been a cost center: spend money to block bad traffic. KYA flips the model: spend money to allow good traffic that generates revenue.
Merchants using F5 can now:
- Block malicious bots (existing capability)
- Identify verified AI agents (KYA integration)
- Allow agents to browse, research, and purchase (policy-based)
- Track agent-driven revenue separately from human-driven revenue (analytics)
This creates a new traffic tier: agent commerce. Not human, not bot — something in between that carries identity, intent, and payment capability.
Availability
The Skyfire integration with F5 Distributed Cloud Bot Defense ships April 30, 2026. F5 customers enable it through the Distributed Cloud Console — no re-platforming, no new infrastructure.
Bottom Line
F5 × Skyfire is the first major CDN/edge security provider to build agent identity into bot defense. It won’t be the last. As AI agents become a meaningful percentage of internet traffic, every WAF, CDN, and bot defense vendor will need an answer to “is this a bot or an agent?”
KYA’s use of standard JWTs is the smart bet — it works with existing OAuth2 infrastructure, doesn’t require merchants to adopt new protocols, and can be verified at the edge without round-trips to identity providers.
The RSAC 2026 pre-wave keeps accelerating. Agent security is no longer just about protecting agents from threats — it’s about giving agents credentials to participate in the economy.