SKILL.md, the King of Shadow AI
Why Agent Skills, as currently architected, are becoming the single biggest accelerator of Shadow AI in the enterprise. And what to do about it.
Read article →AI subscriptions got your teams access. I deploy inside Claude, ChatGPT, or Copilot by instantly adopting the exact roles each department needs. Not one more tool to open.
See me at work2 minutes, no form.
Finance, legal, support, supply chain. I arrive with the blueprint for each function, so you configure a tailored role in three clicks, and nobody re-explains their job to the AI every single time.
Which roles are live, who is using them, and how many hours of work they gave back. One click, no spreadsheet archaeology.
I bundle roles from SKILL.md and scoped MCP tools, pulled on demand by any agent. Architect once, run on 100+ models, rebuild nothing.
I take on a role, not a free pass. I reach only what you granted it, and every task carries the name of the person who sent me.
Visit the Trust CenterSomething else on your mind? My team replies within 24h.
Most organisations have bought Claude, ChatGPT or Copilot licences and watched adoption stall, because every employee has to re-explain their job to the AI each time. skilder fixes that. It is a registry and control plane that turns your know-how into governed roles, one per seat on the org chart, assembled from Agent Skills and scoped MCP tools. Employees keep the AI client they already have; the role arrives inside it. skilder adds no new tool to open.
A Claude Skill is a single SKILL.md file: instructions, optional scripts, reference content. It describes one capability and is portable across any agent that reads the standard. A role is the work one seat actually does: it bundles several skills with the MCP tools that seat may reach, its context and its permissions, and serves them on demand to any MCP-compatible agent. The skill is the capability; the role is how that capability runs in your company, versioned, access-controlled and traced.
No, and that is by design. skilder ships blueprints for 60+ functions, from finance to legal to support, and each one is configured to your playbooks, systems and vocabulary before it goes live. Configuring a role from a blueprint takes a few clicks in a visual interface, not code. Engineers are only needed when you want to expose a custom internal system as an MCP server.
A role is the permission boundary: it reaches only the connectors granted to it centrally, those grants are revocable, and they never widen at runtime. Access is governed by SSO and role-based access control, with tenant isolation between workspaces. Every task is logged and attributed to the person who sent it; skilder keeps that execution trace, and you can export or delete it. Hosting is sovereign in Switzerland or the EU, with on-premise available, and LLM providers run under zero-data-retention terms, so your data never trains a public model.
There is a free plan with usage included, so a builder can configure and run a first role before talking to anyone. Paid plans scale with usage, meaning executions, connected tools and governed users, and add enterprise SSO, advanced audit and dedicated support. skilder does not charge per seat, and authoring roles is always free: you pay for what your organisation actually runs.
skilder Factory is the fast lane. The skilder team works beside yours for two weeks: your first roles configured, SSO and permissions set, everything tested on real work, then the keys handed over. Two ways to work together: delegate the build, or have your own champions backed while they ship. The self-serve path stays free and is enough to configure a first role. Durations, scope and prices are on the Factory page.
Start with one team. Spin up their first role in 2 minutes.
See me at work
Field notes on Agent Skills, MCP and governed AI: what holds up in real deployments, and what quietly does not.
Why Agent Skills, as currently architected, are becoming the single biggest accelerator of Shadow AI in the enterprise. And what to do about it.
Read article →Skilder has been awarded a CHF 20,000 Digital Grant by the Foundation for Innovation and Technology (FIT), the Vaud-based foundation that supports early-stage innovative companies in the canton.
Read article →Shadow AI is the unmanaged use of generative AI tools by employees. Definition, real-world risks, and how to respond without killing productivity.
Read article →