The conducting · Skills and agents
A Mind does not only answer what you ask it. It also runs things: procedures it has learned, agents that act while you are doing something else.
01
A skill is a procedure your Mind knows: the steps for closing the books, drafting a client update, checking a lease against the RRA rules. Once it exists, any agent working against your Mind can use it. No re-explaining the same process to a new agent every time one shows up. Write it once, and it becomes part of what your Mind knows how to do.
02
Anyone can propose a skill. That includes your Mind itself, when it notices you doing the same thing three times and offers to write it down. A draft sits there until you look at it. Only you can approve it and merge it in. That act is the actual security boundary, not a separate approval layer bolted on top. Nothing runs against your Mind until you have read it and said yes.
There is a reason you will not find a catalog of skills written by other people, ready to install without reading. We looked at what happens when other systems open that door. Skills from strangers, running with real access, and no one reading the code before it runs. We decided against it. Every skill your Mind runs was drafted by you, proposed by your Mind, or written in code you chose yourself. That is not a gap. It is the boundary holding.
03
An agent is a process that acts for you. It has its own name, its own permissions, and its own budget for what it can spend. It does not inherit everything you can see. You decide what an agent may touch when you set it up, and it stays inside that boundary. Every action it takes writes a receipt, the same as anything else your Mind does. If you want it stopped, one action ends it: access revoked, budget halted, together.
04
One agent ships with your Mind right now: the morning brief. Every morning, on a schedule you set, it reads what changed since yesterday and assembles a short summary. By the time you open anything else, it is waiting for you. It uses the same governed access as any question you would ask by hand. What it shows you is exactly what the gate allowed it to see.
Agents can run on a schedule, like this one. They can run on an event, such as a document changing. Or you can simply run one by hand. This is not a demo. It runs, and it has been running since before this page was written.
05
A Mind does not stay sealed off from the rest of your business forever. Connections are how it reaches outward: an inbox, a calendar, a piece of software you already run. Each comes in through the same gate as everything else, so every read and every write still gets a receipt. We are still building this out, carefully, one connection at a time. More on this soon.
I did not want an agent that could act as me without a name of its own. If something runs at six in the morning while I am asleep, I want a name to point to and a receipt to check, not a shrug that says the system did it.
ANTONY · FOUNDER
FILED
running since 2026-07-11 · skills and agents shipped, including the morning brief · Connections is still being built
NEXT Yours, provably Ownership, keys, and why hosting is not custody →The founding cohort is small on purpose. Each Mind is brought up personally.
MENSARA.AI · SKILLS AND AGENTS · MACHINE SURFACE · TEXT/MARKDOWN
# Skills and agents > A skill is a written procedure any agent running against a Mind can use. Anyone, including the Mind itself, can draft one; only the owner can ship one, and that act of approval is the security boundary. There is no open catalog of skills from other people. An agent is a named process with its own permissions and budget. Every action it takes is receipted, and it can be stopped in one move. The morning brief is a shipped, scheduled agent that assembles a daily summary through the same governed access as any other read. Canonical: https://mensara.ai/how-it-works/skills-and-agents ## A skill is a way of doing something, written down A skill is a procedure your Mind knows: the steps for closing the books, drafting a client update, checking a lease against the RRA rules. Once it exists, any agent working against your Mind can use it. No re-explaining the same process to a new agent every time one shows up. Write it once, and it becomes part of what your Mind knows how to do. ## Anyone can draft. Only you can ship. Anyone can propose a skill. That includes your Mind itself, when it notices you doing the same thing three times and offers to write it down. A draft sits there until you look at it. Only you can approve it and merge it in. That act is the actual security boundary, not a separate approval layer bolted on top. Nothing runs against your Mind until you have read it and said yes. There is a reason you will not find a catalog of skills written by other people, ready to install without reading. We looked at what happens when other systems open that door. Skills from strangers, running with real access, and no one reading the code before it runs. We decided against it. Every skill your Mind runs was drafted by you, proposed by your Mind, or written in code you chose yourself. That is not a gap. It is the boundary holding. ## An agent runs under its own name An agent is a process that acts for you. It has its own name, its own permissions, and its own budget for what it can spend. It does not inherit everything you can see. You decide what an agent may touch when you set it up, and it stays inside that boundary. Every action it takes writes a receipt, the same as anything else your Mind does. If you want it stopped, one action ends it: access revoked, budget halted, together. ## The one running today: the morning brief One agent ships with your Mind right now: the morning brief. Every morning, on a schedule you set, it reads what changed since yesterday and assembles a short summary. By the time you open anything else, it is waiting for you. It uses the same governed access as any question you would ask by hand. What it shows you is exactly what the gate allowed it to see. Agents can run on a schedule, like this one. They can run on an event, such as a document changing. Or you can simply run one by hand. This is not a demo. It runs, and it has been running since before this page was written. ## What it connects to A Mind does not stay sealed off from the rest of your business forever. Connections are how it reaches outward: an inbox, a calendar, a piece of software you already run. Each comes in through the same gate as everything else, so every read and every write still gets a receipt. We are still building this out, carefully, one connection at a time. More on this soon. **Status:** running since 2026-07-11 · skills and agents shipped, including the morning brief · Connections is still being built ## Prompt ```prompt Ask your Mind to list every skill currently shipped and every agent currently running, including who or what drafted each skill and who approved it. For the morning brief specifically, ask when it last ran, what it read, and whether its most recent receipt verifies. ``` ## Related - [Receipts](https://mensara.ai/how-it-works/receipts) - [Asking](https://mensara.ai/how-it-works/asking) - [Yours, provably](https://mensara.ai/how-it-works/ownership)