A weekly client update looks simple until you have to explain what belongs in it. What counts as finished? Which dates are agreed? What still needs a decision?
An AI agent skill gives the agent a reusable method for that work, with instructions and supporting files. d5s is a no-code AI agent builder: you write the method once in a workspace skill and choose which agents use it.
Here is how to turn a client-update process into a skill your team can use again.
Getting a first draft quickly is useful. Having to explain the process again and correct the same mistakes takes some of that value away. Workday’s January 2026 survey found that nearly 40% of AI time savings went into corrections, rewriting and verification. In Dropbox’s October 2026 survey of 219 AI conference attendees, respondents added context an average of 4.3 more times after their first prompt. This is where a shared skill can help. In d5s, you keep the instructions, examples, template and checks together, so the agent can use the same method each time. You still review the result, but you have less to explain from scratch.
What you need
Choose one recurring job and gather an approved example, a blank template, and the source material. Remove confidential details from examples you do not need to share.
A workspace Owner or Admin can add and edit skills in d5s. Members can read them. You can write a skill in the editor or upload an existing one.
Step 1: Describe the finished work
Give the agent a result you can check.
Start with the sections your team needs. For a client update, that might be completed work, changes to the plan, decisions needed, and next steps.
Prepare a weekly client update from the project records. Link factual claims to their sources. Keep proposed dates separate from agreed dates. Mark missing owners as unconfirmed. Return a draft for review.
The client name and reporting dates change each week. Put those in the task request. Keep the method in the skill.
Step 2: Add the rules and examples
Save the details you would otherwise explain every time.
Tell the agent what to do when records disagree. Explain which internal details to leave out. Give it a blank template and an approved sample, so it can see what you mean.
A template sets the shape of the update. The sample shows the level of detail. The instructions explain the choices between them, so the agent has more than an empty document to fill.
In d5s, these files sit with the skill's instructions. The main file is called SKILL.md. Its description tells the agent when to load the skill.
Step 3: Add the skill in the no-code AI agent builder
Give the right agents the same method.
Open Skills in the workspace sidebar. Select Add → Write a skill. Name it client-weekly-update and describe when to use it. Add the instructions and files.
For an existing skill, choose Add → Upload skills. You can upload a skill folder, a single SKILL.md file, or a supported archive.
A new workspace skill reaches every agent automatically. Open Manage agent → Skills to remove it from agents that should not use it. Skills you remove stay removed when more workspace skills are added.
For a project chat, open Context → Skills and turn the skill on there. See the skills guide for all options.
Step 4: Check the first result
Test the parts that need judgment.
Give the agent a small set of sample records. Include one completed task, one proposed date, and one task with no owner. Ask it to use the skill and return the update for review.
Read the draft beside the records. Check that the proposed date has not become a promise and the missing owner has not become a guess.
Step 5: Keep useful corrections
Update the method when the correction should apply next time.
“Put decisions first in every update” belongs in the skill. “Leave out this cancelled workshop” belongs in the current task.
A workspace Owner or Admin can edit the skill. Agents use the newest version on their list. Replacing a skill through upload saves a new version and keeps the previous one in View history.
Check another sample after a change. Then use the same process for another recurring job from our AI agent use cases.
Questions and answers
What is the difference between an AI agent and a skill?
An agent has a role and carries out work. A skill supplies a reusable method for a particular task. In d5s, an agent uses the skills on its own Skills list. For AI workflow automation, add an automation that runs the job on a schedule.
Do I need to write code to create a skill?
No. You can write instructions in the d5s skill editor and add supporting files. An uploaded skill needs a `SKILL.md` file with a name and description.
Does a skill give an agent access to my apps?
No. A skill describes how to do the work. Connections and permissions control access to apps and actions. Review those separately before asking the agent to read or change data.
