Once you know how to create a skill, the next step is learning how to write instructions that actually work and how to build habits that make skills useful across your whole team.
Using a Skill in the AI Assistant
Open the AI Assistant.
Click the + button in the chat input.
Select Skills from the menu and choose the skill you want to apply.
📌 Type / in the chat input to search for and activate a skill without leaving the keyboard, or @mention a skill by name directly in your message to apply it inline.
The skill stays active for that session: combine it with your message for best results.
📌 You can apply multiple skills in the same conversation: just add them one by one using the + button or by @mentioning them.
Sharing a Skill With an Agent
To make an agent follow a skill's instructions every time it runs, share the skill with that agent.
Open the skill and click the share icon.
Add the agent and set access to Can view.
Save.
The agent will apply the skill's instructions automatically. You can share multiple skills with a single agent.
To remove a skill from an agent, open the share dialog and remove the agent. The skill itself is not affected.
You can also add skills to an agent directly from the agent's settings.
👉 Find out more in Setting Up and Managing Agents.
Writing Instructions That Work
Instructions are the core of a skill. The more specific they are, the more consistent the output.
Be explicit about format
Instead of "Summarize this", write "Summarize in three bullet points: key decisions, action items, and open questions."
Format instructions remove ambiguity and make outputs easier to reuse.
Set the tone
If the output has an audience (e.g. a client, a stakeholder, a team) say so. "Write in a confident, professional tone. No jargon, no filler."
Add constraints
Constraints help keep outputs focused. "Keep responses under 150 words." "Always respond in the language of the input." "Never suggest changes to scope without flagging the budget impact."
Think in workflows, not just prompts
The best skills encode a multi-step process, not just a single instruction. If you find yourself repeating the same sequence of steps in chat, that's a skill waiting to be written.
Referencing Data in Instructions
You can reference Productive data directly inside your instructions using @ mentions — people, projects, tasks, deals, and other objects. This is useful when a skill is always tied to a specific context, like a particular project, doc, or team member.
If you or an agent have a connector linked (like Amplitude, HubSpot, or any other integration or app), you can also reference it in the instructions or chat:
"Go to Amplitude, open the Productive App project, and pull the most recent event data for the last 7 days."
The agent or AI Assistant will follow those instructions as part of its workflow, pulling in external data without you having to do it manually each time.
Practical Examples
Client status update
"Draft a project status update for a client. Include: progress since the last update, what's coming next, and any active risks. Tone: professional and reassuring. Max 200 words."
Activate when drafting client comms: consistent format, every time.
Meeting recap
"Read this meeting transcript and extract: key decisions, action items (with owner and deadline if mentioned), and open questions. Open a new doc and use bullet points to structure it."
Access a meeting recording, activate the skill, get a structured recap in seconds.
Budget risk review
"When reviewing budget data, flag any projects where margin is below 20%. For each, show the variance in hours and cost, and flag projects where burn rate suggests they'll exceed the budget before the end date. End with a recommended next action."
Share this with a budget-focused agent, or activate it manually when reviewing financials.
Onboarding checklist generator
"Given a new hire's role and department, generate a two-week onboarding checklist in an Onboarding task list. Include: day-one account setup, tools and access needed, key people to meet in week one, and one goal to complete by end of week two. Adapt the list to the role."
A reliable starting point every time someone joins the team.
Building a Skill Library for Your Team
Skills become more valuable when they're shared. A few patterns that work well:
Start with your most repeated workflows: if someone on the team does the same thing in chat every week, turn it into a skill and share it
Use descriptions well: a clear description helps teammates find and understand a skill without having to open it; treat it like a label on a tool
Iterate openly: when a shared skill isn't producing great output, update the instructions and let the team know what changed; skills improve through use
Start from templates: Productive's built-in templates cover common use cases across project management, resourcing, budgeting, and time tracking; customize them rather than starting from scratch
Related Articles
Not sure where to go next? These articles cover everything around skills and AI in Productive.
👉 What Are Skills?: understand what skills are and how they work before building your own
👉 How to Create and Manage Skills: step-by-step instructions for creating, editing, sharing, and deleting skills
👉 Agents: set up autonomous agents that run tasks on your behalf using assigned skills
👉 AI Assistant: learn how to chat with Productive's AI Assistant and get things done faster