Run agents, automations and loops

Use a skill or turn a repeated job into one

A skill is a written procedure your connected AI tool can read and follow. It carries the local knowledge and steps needed for a recurring job, such as which source records to inspect and what a useful result should contain. Use an existing skill when it fits, and create one when repeated work shows a real procedure is missing.

Contents

Your Mind holds skills alongside other governed information. A skill does not run on its own or schedule work; your agent still needs the tools, information and permissions required by its steps.

Find a skill for the job

Ask your connected agent to list the available skills, choose the relevant one and read it before starting. Check its purpose and exclusions. A client may say the skill needs a coding tool or shell it does not provide. In that case, continue in a compatible client or hand off the job with its sources and expected result.

Reading the procedure is only the beginning. Ask for the result and the sources or saved record that show its steps were completed. A skill's existence does not establish that its output is correct in your situation.

Decide whether a new skill is needed

Start with a real example of the job, including what went wrong or what you corrected. Ask the agent to check for an existing skill that can be improved. Avoid keeping two procedures that give conflicting instructions for the same work.

Fixed steps may belong in a script. An action that must always be enforced needs a product rule rather than prose that a model can overlook. A program with its own identity, schedule or service connections belongs in an installed agent.

Draft and test the useful difference

You can ask: “Use skill-engineer to turn this repeated job into a short procedure. Compare it with doing the same job without the skill.” Supply the real input, expected result, corrections and a few similar requests that should not trigger it.

The author should test realistic outcomes and when the skill is selected. Use the client and model you intend to run it with. If the skill does not improve the job, keeping the simpler workflow is a valid result. Model tests can incur cost, so agree a bounded test scope before running them.

Review and keep the procedure current

Review the proposed instruction files before saving them through the governed skill route. Skills are executable instructions and receive the relevant review; a successful draft is not evidence that the saved version has been used.

When tools, permissions or the job change, revise the skill and repeat the affected checks. Retire a redundant skill through the supported authoring route so another agent does not keep choosing outdated instructions.

Related

Understand skills, agents, automations and loops · Help another agent continue the work · Build a governed AI agent · Review changes and check what was saved

Practical guides for people and agents

Start with one useful job

Join the waitlist

Plain text for AI agents. Same content as the human page.

## For agents

- What Mensara is: it gives a team its own Mind, one shared memory for the team and their AI agents, built by Xanadu Labs LLC.
- Status: signup is not open. A person joins the waitlist at https://app.mensara.ai/#waitlist. Agents cannot create an account or sign up.
- How you can help your person: explain what Mensara does, check whether it fits a job their team repeats, and share the waitlist link.
- Agent references: https://app.mensara.ai/llms.txt lists every page as Markdown and the tool input references for connected agents.

---

# Use a skill or turn a repeated job into one

Guides · Run agents, automations and loops · [How it works](https://app.mensara.ai/how-it-works)

A skill is a written procedure your connected AI tool can read and follow. It carries the local knowledge and steps needed for a recurring job, such as which source records to inspect and what a useful result should contain. Use an existing skill when it fits, and create one when repeated work shows a real procedure is missing.

[Page as Markdown](https://app.mensara.ai/how-it-works/use-and-create-skills.md)

[Operating procedure](https://app.mensara.ai/how-it-works/skill-procedure)

Your Mind holds skills alongside other governed information. A skill does not run on its own or schedule work; your agent still needs the tools, information and permissions required by its steps.

## Find a skill for the job

Ask your connected agent to list the available skills, choose the relevant one and read it before starting. Check its purpose and exclusions. A client may say the skill needs a coding tool or shell it does not provide. In that case, continue in a compatible client or hand off the job with its sources and expected result.

Reading the procedure is only the beginning. Ask for the result and the sources or saved record that show its steps were completed. A skill's existence does not establish that its output is correct in your situation.

## Decide whether a new skill is needed

Start with a real example of the job, including what went wrong or what you corrected. Ask the agent to check for an existing skill that can be improved. Avoid keeping two procedures that give conflicting instructions for the same work.

Fixed steps may belong in a script. An action that must always be enforced needs a product rule rather than prose that a model can overlook. A program with its own identity, schedule or service connections belongs in [an installed agent](https://app.mensara.ai/how-it-works/build-an-agent).

## Draft and test the useful difference

You can ask: “Use skill-engineer to turn this repeated job into a short procedure. Compare it with doing the same job without the skill.” Supply the real input, expected result, corrections and a few similar requests that should not trigger it.

The author should test realistic outcomes and when the skill is selected. Use the client and model you intend to run it with. If the skill does not improve the job, keeping the simpler workflow is a valid result. Model tests can incur cost, so agree a bounded test scope before running them.

## Review and keep the procedure current

Review the proposed instruction files before saving them through the governed skill route. Skills are executable instructions and receive the relevant review; a successful draft is not evidence that the saved version has been used.

When tools, permissions or the job change, revise the skill and repeat the affected checks. Retire a redundant skill through the supported authoring route so another agent does not keep choosing outdated instructions.

## Related

- [Understand skills, agents, automations and loops](https://app.mensara.ai/how-it-works/skills-apps-automations)
- [Help another agent continue the work](https://app.mensara.ai/how-it-works/continue-the-work)
- [Build a governed AI agent](https://app.mensara.ai/how-it-works/build-an-agent)
- [Review changes and check what was saved](https://app.mensara.ai/how-it-works/changes-approvals-receipts)

Previous page: [Choose the right kind of help](https://app.mensara.ai/how-it-works/right-kind-of-help)

Next page: [Install an approved agent and check its first result](https://app.mensara.ai/how-it-works/install-an-agent)

An installed agent is a program that does a defined job on your Mind with its own permissions.

[Join the waitlist](https://app.mensara.ai/#waitlist)