​TrueFoundry Agent Harness Hackathon

​AI agents are easy to prototype and much harder to run in production. Once an agent touches real users, tools, data, budgets, and internal systems, the work becomes less about a clever prompt and more about routing, permissions, tracing, retries, evals, guardrails, cost controls, and deployment.

​Join TrueFoundry and HackerSquad for a full-day Agent Harness hackathon focused on building production-minded agentic applications. The goal is simple: take an agent use case from idea or prototype toward something that can be observed, governed, evaluated, and operated.

​TrueFoundry's AI Gateway, Agent Gateway, and MCP Gateway will frame the day. Builders will work on practical agent harness patterns such as model routing, controlled tool access, MCP workflows, observability, eval loops, fallback behavior, budget controls, governance, and deployment paths.

​This is a technical builder event, not a passive product pitch. Bring a laptop, a strong use case, and a willingness to build under realistic production constraints.

​What You Can Build

• ​An agent workflow with model routing, retries, and fallback behavior

• ​A tool-using agent with controlled MCP or API access

• ​An observability layer for traces, tool calls, latency, usage, cost, and failures

• ​An eval or test harness for agent behavior across repeated runs

• ​A governance pattern for permissions, approvals, logging, blocking, or escalation

• ​A deployment-minded agent application that shows how it would run inside a real team or platform environment

​Why This Gets Hard in Production

• ​Agent systems need to coordinate models, tools, permissions, memory, context, retries, and fallbacks without becoming brittle one-off glue code.

• ​Teams need traces, audit logs, usage visibility, cost attribution, and eval loops so they can understand what an agent did, why it did it, and what changed between runs.

• ​Scaling agents means managing quotas, budgets, latency, provider failures, tool access, and security policies across teams.

• ​Guardrails need to become runtime controls: clear rules for what data an agent can see, which tools it can call, which actions require human approval, and how unsafe requests or outputs get blocked, logged, or escalated.

• ​Deployment gets harder when agents need to connect to internal systems, respect governance rules, and stay reliable as usage grows.

​Who Should Apply

• ​Platform, infrastructure, and backend engineers building or operating AI agent systems

• ​AI engineers and product engineers working with LLMs, tool calling, MCP, evals, or agent workflows

• ​Technical founders moving agent prototypes toward production

• ​Security, DevOps, and ML platform practitioners thinking about governance, observability, and cost control

• ​Builders with a concrete agent use case and enough technical depth to build during the day

​Format

• ​Doors, check-in, coffee, and setup

• ​Welcome from HackerSquad / The Builder Collective and TrueFoundry

• ​TrueFoundry Agent Harness framing

• ​Challenge brief, build tracks, and judging rubric

• ​Focused build sprint with technical support available

• ​Project submissions and proof packages

• ​Selected demos, judging, awards, and closing remarks

• ​Follow-up conversations and networking

​Bring

• ​A laptop

• ​A concrete agent use case, workflow, prototype, or infrastructure problem

• ​Questions about model gateways, MCP tools, observability, evals, cost controls, deployment, or agent governance

• ​Any teammates you plan to build with

​Admission is free and approval-based because space is limited and the event is designed for technical builders working on relevant agent infrastructure and application problems. The exact South Bay venue and arrival details will be shared with approved attendees once the venue path is confirmed.

​When: Saturday, September 19, 2026 | 9:30 AM - 8:00 PM PT

​Where: South Bay, Location TBD.