# Skill Manager: New Capabilities for Your AI Agents, Tested > The Skill Manager is the builder of AI agent skills in the Ai1 platform by MyZone AI: you describe the ability you want and approve its plan, then it builds the skill, keeps any key in secure storage on your Ai1 server and tests the skill end to end before your agents use it. Canonical page: https://myzone.ai/pages/agents/ai-skill-manager-agent Part of Ai1, by MyZone AI. Book a Skill Manager walkthrough: https://calendly.com/d/ct6h-tcy-8qf/ai1-demo?a1=Skill%20Manager&utm_source=myzone.ai&utm_medium=agent-page&utm_content=ai-skill-manager-agent-final Last updated: 2026-10-05. Reviewed by the MyZone AI team. New abilities for your agents, planned with you and tested before use. ## At a glance - **You tell it:** What your agents should be able to do, in plain language - **Before building:** A plan with the skill's name, purpose and any script or login, for your approval - **Keys and logins:** Entered by a person through a secure form and kept in secure storage, never in chat - **Testing:** Run end to end, with an independent review on anything non-trivial - **Existing skills:** Fixed with a record of what changed, or retired when no longer needed - **What it never does:** Edit the platform's built-in tools, which are read-only ## What the Skill Manager does - **Builds a new skill from a plain request:** You say what an agent should be able to do, such as pull data from a service or automate a step, and it designs and builds that skill from scratch. - **Shares the plan before building:** Works out the skill's name, purpose and whether it needs scripts or a login, then shows you the plan to approve before any building starts. - **Keeps credentials safe:** When a skill needs a key for an outside service, a person enters it once through a secure form. It is stored on your Ai1 server and never hardcoded or exposed. - **Tests end to end:** Runs every new skill in full, confirms it finishes cleanly without leaving anything running, and asks for an independent review on anything non-trivial. - **Fixes and updates existing skills:** When a skill breaks or needs a change, it finds the cause, edits the right skill, records what changed and checks the fix works. - **Retires skills you no longer need:** Keeps your set of skills tidy by removing the ones that have outlived their use. ## How the Skill Manager differs from a one-off script A one-off script can end up holding a key in its code and in use before anyone has run it in full. The Skill Manager shows you the plan before it builds anything, has a person enter any key through a secure form and runs every new skill end to end, with an independent review on anything non-trivial. It also fixes and retires skills, and never edits the platform's read-only built-in tools. How this differs from the Agent Manager (https://myzone.ai/pages/agents/ai-agent-builder-agent): the Agent Manager creates, upgrades and renames the agents themselves, while the Skill Manager builds and tests the skills and tools those agents use. ## How it works 1. **Describe the ability you want** (Whenever you need a new ability): You explain what you want your agents to be able to do, in plain language. 2. **Approve the plan** (you approve) (Before any building): It sets out the skill's name, what it does and whether it needs scripts or a login, and waits for your go-ahead before building. 3. **Build** (After your go-ahead): It writes any scripts the skill needs, has the login entered through the secure form and creates the skill for your agents. 4. **Test** (Before it is called done): It runs the skill end to end, checks it closes cleanly and reviews the logic before calling the job done. 5. **Ready or handed on** (Once its tests pass): The skill is live for your agents, or passed to the separate publishing step if it is meant for your wider organisation. ## When to use it - **You want an agent to pull information from another tool you use:** It builds a skill that reads from that tool, such as a weekly summary of new leads from your CRM, and tests it before your agent relies on it. - **A skill that used to work has started failing:** It diagnoses the fault, fixes the right skill and confirms it runs without errors. - **You need a repetitive file task handled:** It builds a skill for jobs like resizing images before upload, so agents can do it the same way each time. - **You want to extend a built-in platform tool:** Built-in tools stay read-only, so it offers to build a separate copy that is fully yours to change. ## What you get - A skill plan for your approval before any building starts - New skills built, tested end to end and ready for your agents - Credentials stored securely on your Ai1 server, never in code or chat - Fixes to existing skills with a record of what changed - Finished skills handed to the publishing step when they are meant for your wider organisation ## Example: weekly leads summary skill Example with a fictional company. Names, people and figures are invented to show the agent's output. Any resemblance to a real company or person is unintended. What it was asked: Build a skill that pulls this week's new leads from our CRM and gives our sales agent a short Monday summary it can share with the team. The Skill Manager turned the request into a plan, asked two questions and waited for the operations manager's approval in chat before building. It wrote the skill instructions and one read-only script, had a person enter the CRM key through the secure form, then ran six end-to-end tests and asked a separate reviewing agent to check the logic. It did not set up the Monday schedule or publish the skill to the wider organisation; both are separate steps a person decides on. Run date: 30 September 2026. ### What it found - The CRM sends leads 25 at a time. The first draft missed 26 of 51 records in a busy week; the skill now reads every page. - Four customers filled in the web form twice. Merging by email or phone gives 47 real leads, and 3 with no contact details are listed for a follow-up. - All 6 tests pass and the longest run takes 3.8 seconds. The key stays in secure storage and never appears in code, chat or output. ## Guardrails - Nothing is built until you approve the plan. - Keys and logins are entered by a person through a secure form, never pasted into chat, and are never written into a skill or shown in its output. - A skill is not called done until it has run end to end and finished cleanly. - It never edits the platform's built-in tools, which are read-only. It only changes skills that are yours to change. ## Frequently asked questions ### How do AI agents connect to the other tools a business uses? Through a skill that reads from or works with that tool. The Skill Manager plans the skill with you, a person enters the tool's key once through a secure form, never in chat, and the key stays in secure storage on your Ai1 server. It is fetched only when the skill runs, so agents can use it without anyone seeing it. You only hand over a key when a skill connects to an outside service. ### How is a new AI skill tested before agents rely on it? The Skill Manager runs it end to end, confirms it finishes cleanly without leaving anything running and asks for an independent review on anything non-trivial. A skill is not called done until it passes, so bugs are caught before your live agents meet them. Sharing a skill across your whole organisation is a separate publishing step, which audits it and decides on the release. ### Can I add new agent capabilities without writing code? Yes. You say what an agent should be able to do, such as pull data from a service or process files, and approve the plan. The Skill Manager writes any scripts the skill needs and builds it, so nobody on your team has to write code. A skill is a reusable set of instructions, sometimes with a script, that an agent picks up to do one specific job, such as reading a spreadsheet, sending a notification or calling an outside service. ### How is the Skill Manager different from the Agent Manager and Recipe Manager? The Agent Manager creates the agents. The Recipe Manager writes the step-by-step procedures they follow. The Skill Manager builds the individual tools and abilities agents use inside those procedures. It cannot change the tools that come with the platform: built-in tools are read-only. If you want one extended, it offers to build a separate, editable version instead. ## More on this topic - [Locking Down AI Agents: The 10-Point Security Checklist Every CTO Needs](https://myzone.ai/pages/blog/ai-agent-security): AI agents are powerful,and vulnerable. Learn the 10 critical security steps to lock down your AI agents before they become your biggest liability. - [Multi-Agent Development: Why the Future of Software Is Built by AI Teams](https://myzone.ai/pages/blog/multi-agent-development): Discover how multi-agent AI systems are transforming software development. Teams of specialised AI agents build products 10x faster at lower cost. ## About Ai1 Ai1 is the AI operations platform by MyZone AI, where each client runs on its own private server. The Skill Manager is included on every Ai1 level, including Developer Core, with no per-agent charge. It works alongside the other Ai1 agents on your account. Pricing: https://myzone.ai/pages/services/ai1-pricing. Security: https://myzone.ai/pages/security.