
I thought everyone was building tools for themselves
At ContentCon, I realised building tools for yourself still feels unfamiliar to many developers. Coding agents help, but finding a starting point matters too.
I’m Tim Benniks, Developer Experience and Product Leader at Contentstack. I build developer tools and write about AI engineering, composable architecture, and what makes software useful.

The journal
Essays on developer experience, AI-accelerated engineering, and the platforms that hold it all up.

At ContentCon, I realised building tools for yourself still feels unfamiliar to many developers. Coding agents help, but finding a starting point matters too.

This article explores an unexpected result from evaluating Contentstack’s MCP Profile Hub, where increasing a model’s reasoning effort made straightforward CMS agent tasks slower and less reliable. While enriching tool definitions with account-specific context clearly reduced tool calls, latency, and reasoning tokens, turning up the reasoning setting caused the model to distrust simple, correct tool responses and invent extra work. A control task listing global fields showed high-effort runs repeatedly ignoring a perfect API answer and exhausting their turn budget. The piece argues that reasoning effort must match the shape of the work, that bounded API operations often benefit from lower reasoning, and that agent evaluations should measure behaviour—calls, latency, traces, and early stopping—rather than abstract “intelligence” alone.

How the new MCP Skills Extension works, from skill discovery and on-demand loading to content verification, workflow guidance, and managed profiles.
The workbench
Agentlint, platform SDKs, local-first Mac apps, and the occasional weekend experiment. Real code, real trade-offs, work in progress.
Audit whether AI agents can find, understand, and use your website with Agentlint, a deterministic CLI scanner that backs its findings with evidence.
Active · @timbenniks/agentlint
Build with one TypeScript SDK for Contentstack, covering the Content Management API, OAuth, webhooks, image transforms, and generative AI.
Active · @timbenniks/contentstack-platform-sdk
The Nuxt module I wish Contentstack shipped: typed composables, live preview, visual builder, personalization, and image optimization in one install.
OSS · nuxt-contentstack on npm
Content types as TypeScript you can import, compile, diff, document, type-generate, and safely promote, so the UI is no longer the source of truth.
OSS · @timbenniks/contentstack-stacksmith
Expose published Contentstack content to browser agents through WebMCP, with framework-agnostic TypeScript tools and adapters for React and Next.js.
Experimental · @timbenniks/contentstack-webmcp
A hub of Contentstack agent skills with role-based access control. Agents only reach permitted data, and product managers can propose new skills.
Active · Internal + open contributions
The short version

Twenty years building digital platforms for global brands. Now leading Developer Experience at Contentstack, shaping the future of developer-facing products and AI-driven workflows.
Tim Benniks is a Developer Experience and Product leader at Contentstack, where he drives DX, Developer Hub & Marketplace, MCP, and Agent Skills. Tim is a frequent speaker, content creator, ambassador for Cloudinary, Supabase, Nuxt, Algolia and Co-chair of the Enterprise AI Agents Workgroup at the MACH Alliance.
When output is cheap, taste is everything.
AI makes it easier to build. Choosing what deserves to exist still takes judgment, curiosity, and a rested mind.
On camera
Workshops, demos, and conversations from the YouTube channel and Contentstack streams.

Whether you’re new to the concept of AI agents or already exploring ways to scale your content operations, this session will walk through the fundamentals of Agent OS, and how you can get your teams ready ahead of GA.

Learn how to activate Visual Builder, no matter your stack

See how Contentstack Edge allows you to reimagine the possibilities of digital experience by seamlessly integrating intelligent content management with real-time first-party data activation. Experience how brands can…
Stages
ContentCon Amsterdam 2026
A Developer Labs session with Tim Benniks and John Kelly covering Pulsar, Agentlint, and WebMCP. Listed as Part 1 of the developer lab in the ContentCon Europe agenda.
Topics: AI engineering · Developer experience
Format: talk
Sources checked 2026-09-22
ContentCon Amsterdam 2026
A hands-on certification workshop for developers and marketers using Contentstack Agent OS. Covers context engineering, prompting, MCP, skills, and AI-driven personalization, with certification based on building and presenting a working agent.
Topics: AI engineering · Personalization · Content ops
Format: workshop
Sources checked 2026-09-22
MACH X Amsterdam 2026
A hands-on lab introducing four agent architecture patterns before participants build a working agent. The agenda includes a foundations session followed by open building time for business and technical attendees.
Topics: AI engineering · Composable architecture
Format: workshop
Sources checked 2026-09-22
vuejs.amsterdam 2026
Revisiting the 2013 Need for Speed Rivals campaign and rebuilding it with modern Vue tools. Compares hand-built routing, state management, animations, and asset loading with Vue Router, Pinia, Vite, and composables.
Topics: Frontend · Developer experience · Craft
Format: talk
Sources checked 2026-09-22
Got a stage, a podcast, or a developer team that needs sharpening? Let's talk.