How to use the Project Management Expert agent skill

What is an agent skill and what does this one do?

An agent skill is a folder of instructions that an AI agent loads so it can handle a certain kind of task in a consistent way. The project-manager skill is written as guidance and principles, not as a tool that connects to your project software.

When the skill is active and you ask for help planning work, the agent applies a project manager's approach to software teams. It covers estimation, risk management, stakeholder alignment and team health, and it reasons about trade-offs such as adding new work in the middle of a sprint.

• Run effective standups focused on blockers and coordination, timeboxed to 15 minutes.
• Facilitate sprint planning by breaking epics into stories with clear acceptance criteria and committing to a realistic sprint goal.
• Run retrospectives using formats such as Start/Stop/Continue, 4Ls or sailboat, and track action items into the next retro.
• Build a RACI matrix (Responsible, Accountable, Consulted, Informed) to clarify decision rights on cross-team work.
• Track velocity over 3-5 sprints for forecasting, and use burndown charts within a sprint and burnup charts for release progress.
• Write stakeholder communication plans that specify audience, frequency, channel and level of detail.
• Maintain a living risk register with probability, impact, owners and mitigation plans.

How do you install the project-manager skill?

You can install the skill in one of two ways. Download the ZIP from this page and unzip it into your personal skills folder, ~/.claude/skills/, or into a project's .claude/skills/ folder if you want the skill available only in that project.

Alternatively, if you use the Claude apps, upload the skill under Settings → Capabilities → Skills.

• Personal install: unzip the download into ~/.claude/skills/ so the project-manager folder sits there.
• Project install: unzip the download into the .claude/skills/ folder inside your project.
• Claude apps: go to Settings → Capabilities → Skills and upload the skill.

How do you trigger the skill with natural requests?

You do not need a special command. Describe the planning problem in plain language and mention the kind of help you want, such as a sprint, a retro, a risk or a stakeholder update. The skill is meant to be used when you ask for help planning work.

Give the agent context: team size, sprint length, what is blocking you, and who the audience is. The more specific the request, the more useful the output.

• "Help me plan next sprint for a team of five engineers. Our goal is to ship the new billing export."
• "Set up a Start/Stop/Continue retro for our team after a rough release."
• "Draft a RACI matrix for the payments and mobile teams on the new checkout project."
• "Create a risk register for our migration to the new database, with owners and mitigations."
• "Write a weekly stakeholder update for executives on our product launch."
• "A stakeholder wants to add a feature mid-sprint. How should we weigh that against the sprint goal?"

What can you ask it to do? Examples and patterns

The skill includes several recurring patterns that you can ask the agent to apply. Each one addresses a common friction point in software delivery.

Scope negotiation: when a new request arrives mid-sprint, the skill advises adding it only if an equivalent item is removed. Dependency mapping: it suggests identifying cross-team dependencies at the start of each planning increment and assigning owners for handoffs. Risk-based sequencing: it suggests scheduling high-uncertainty work early so there is time to adjust. Definition of done: it helps you write a team-agreed checklist, such as code reviewed, tests passing, documentation updated and deployed to staging.

What does the skill need, and what about scripts and network use?

The skill lists no requirements. It does not run scripts and does not use the network. The source contains a single file, SKILL.md, which holds the guidance.

Because it is only instructions, it has no access to your Jira, Asana or other live project tools. The agent cannot read or update your tickets, boards or schedules through this skill. Any data you want considered must be pasted or described in your request.

• Requirements: none listed.
• Scripts: none.
• Network access: none.
• Files: SKILL.md only.

What are the limits and pitfalls to know?

The skill is built around a few clear principles, and it warns against common misuse. It treats estimation as a communication tool rather than a commitment contract, and it advises against equating story points with hours or using velocity as a performance metric.

It is general Agile guidance. It does not certify any process, so it is not a substitute for formal methods such as PRINCE2 or PMP compliance on regulated or government projects.

• Do not turn story points into hours or use velocity to judge people.
• Do not skip retrospectives when the team is busy.
• Do not rely on status meetings alone; talk to individual contributors.
• Do not commit to deadlines without consulting the engineering team.

Tips for getting better results

Treat the output as a starting draft. Adapt ceremonies and artifacts to your team's real needs, since the skill itself says Agile is a mindset rather than a rigid set of rituals.

Bring your own data. Paste in your backlog items, past sprint velocities or the list of known risks so the agent can work from specifics.

• Name the audience for any status update so the level of detail fits.
• Ask for a risk register in a table so owners and mitigations are easy to review.
• Ask the agent to list the assumptions it made, then correct any that are wrong.