Topic starter
02/08/2026 12:16 pm
TL;DR: At an AWS AI Agents Hackathon, more than 250 developers built 50-plus projects around agentic workflows, with MCP, multi-LLM patterns, and in-browser security checks emerging as the dominant themes, according to Semgrep. The real issue is that faster agent-enabled development is expanding identity, tool, and trust boundaries before most governance models are ready.
Editorial analysis by NHI Mgmt Group, based on content published by Semgrep: “What a Hackathon Reveals About AI Agent Trends to Expect in 2026”.
By the numbers:
- Over 250 developers gathered to build 50+ projects at the AWS AI Agents Hackathon.
Key questions
Q: What breaks when AI agents use MCP without stronger governance?
A: Governance breaks when the organisation assumes access happens through a stable, inspectable human workflow.
Q: Why do multi-LLM developer workflows increase the need for deterministic security checks?
A: Because model choice becomes fluid, security cannot rely on the behaviour of any one model or provider.
Q: What are the signs that browser-based AI coding is bypassing security controls?
A: Common signs include generated code moving from browser to repository without being scanned, security feedback appearing only after commit, and developer tools lacking integrated policy checks or inspection points.
Practitioner guidance
- Govern MCP server admission Require a formal review before any MCP server is connected to an agentic development environment.
- Embed deterministic code scanning in the editor Run static analysis inside the IDE or browser session so generated code is checked before it is copied, committed, or deployed.
- Separate model choice from security approval Allow developers to switch between LLMs for task fit, but keep security approval, policy checks, and release gating independent of model selection.
Bottom line: MCP is emerging as the connective tissue for agentic development, which moves trust, access, and inspection decisions into the developer workflow itself.
Explore further
View Full Forum → | NHI Foundation Course → | Our Services → | Read the full analysis →
02/08/2026 1:18 pm
MCP is turning developer tooling into an identity and authorisation problem, not just an integration problem. Once agents can invoke tools, pull context, and interact with environments through a standard protocol, security teams have to govern who or what is allowed to act inside the workflow. That makes the access boundary the control point, not the model selection. Practitioners should evaluate MCP as a delegated-access layer that needs explicit policy, auditability, and revocation paths.
A few things that frame the scale:
- 70% of organisations grant AI systems more access than they would give a human employee performing the exact same job, according to the 2026 Infrastructure Identity Survey.
A question worth separating out:
Q: How should security teams decide whether to allow third-party MCP servers?
A: Treat them like privileged integrations rather than ordinary plugins. Teams should review what systems the server can reach, whether its code has been assessed, how access can be revoked, and what audit evidence will exist if the server is used in a consequential workflow.
👉 Read our full editorial: MCP is becoming the backbone of agentic developer workflows