# AI for Project Management: 31 Claude Skills to Run a Project AI for project management: 31 free Claude skills for the charter, scope, estimates, Gantt chart, critical path, risk register, status report and handover. - Page: https://lantern.meet-polar-bear.com/skills/ai-for-project-management-pack - Download: https://lantern.meet-polar-bear.com/skills/ai-for-project-management-pack.zip - Source: https://github.com/polar-bear-org/claude-skills/tree/main/packs/ai-for-project-management-pack - Price: free ## Install in Claude Code ``` /plugin marketplace add polar-bear-org/claude-skills /plugin install ai-for-project-management-pack@polar-bear-skills ``` ## The 31 skills you get The set follows the project, in eight stages: starting it, stopping scope creep, estimating and scheduling, catching risks early, getting decisions made, running the delivery rhythm, reporting honestly, and closing well. ### 1 · Start the project - **Project Charter** — use when: The project arrived as one sentence, with no sponsor named. Output: Outcome-based objectives with success measures, sponsor and project manager authority by role, high-level scope with named exclusions, milestones, a budget envelope and a sign-off block. - **Kickoff Meeting Agenda** — use when: You want the kickoff to settle scope, owners and change handling. Output: A timed agenda with the hard questions — what done looks like, what is out, who decides, how changes are handled — a roles walkthrough, a pre-read, and the outputs to confirm by the end. - **Project Scope Statement** — use when: Every “small” ask is arguable because scope was never agreed. Output: In-scope deliverables with testable acceptance criteria, named exclusions, constraints and owned assumptions, and a sign-off block. ### 2 · Stop scope creep - **Change Request Form** — use when: Someone says “it's just one small change”. Output: The ask sized on time, cost, scope, quality and risk, the options — accept, defer, swap, reject — with a named decider and a date, and the change log entry. - **MoSCoW Prioritization** — use when: Scope, date and team are fixed and something has to give. Output: A Must, Should, Could and Won't-this-time list with the effort share per category, the drop order if a new Must arrives, and a sign-off line. - **Work Breakdown Structure** — use when: The project is one big blob and nobody can say what done contains. Output: A deliverable tree down to work packages with a 100% rule check, dictionary rows, a proposed owner per package, and a gaps list of forgotten work. ### 3 · Estimate and schedule - **Three-Point Estimate** — use when: Leadership wants a date before anyone has estimated. Output: PERT mean and spread per work package, the forgotten work listed, and a range with a confidence line for the whole project. - **Dependency Map** — use when: Nobody can see what is blocking what. Output: A cross-team and external dependency table, a text diagram of what blocks what, and the at-risk list. - **Critical Path** — use when: Everything is urgent and you need to know which slip moves the end date. Output: The activity network, a forward and backward pass, total float per activity, and the list of delays that move the end date. - **Gantt Chart** — use when: The Gantt got the project approved and has not been touched since. Output: A schedule table and a Mermaid Gantt chart, baseline against current dates, rolling-wave detail for the next period, and change notes. - **Resource Allocation Plan** — use when: Leadership wants twice what the team can deliver. Output: Demand by role and skill per period against capacity, the overloaded and bottleneck roles, levelling options with their date impact, and the trade-off for the sponsor. ### 4 · Catch risks early - **Pre-Mortem Analysis** — use when: Risks only surface after something nearly goes wrong. Output: Grouped failure reasons from the team, the top five turned into owned risks, and the plan changes to make now. - **Risk Register** — use when: The register was built at kickoff and never opened again. Output: Each risk worded as cause, event and effect, likelihood and impact scored on defined scales, an owner, a trigger, a response and a review date. - **RAID Log** — use when: The log died by month three. Output: The top risks from the register, assumptions with a test date, issues with owner and tolerance, dependencies with need-by dates, stale-entry flags, and the weekly review agenda. ### 5 · Get decisions made - **Stakeholder Map** — use when: Decisions stall because nobody is sure who decides. Output: The stakeholder list by role, a power and interest grid built on formal decision rights and how much the project changes each role's work, and a table of who decides what. - **RACI Matrix** — use when: A missed deadline traced back to nobody owning the task. Output: Deliverables against roles with exactly one Accountable per row, gaps and overloaded role columns flagged, and the questions to settle in the room. - **Escalation Email** — use when: Six people are “reviewing” and nobody has a deadline. Output: One named decider asked for one decision by a date, the options with their impact, a recommendation, the default if no answer comes, and a short version for chat. - **Steering Committee Deck** — use when: The steering meeting became an update session. Output: The decisions needed first, then delivery confidence with its reasons, milestones against forecast, top risks, changes for approval, and draft resolutions for the minutes. - **Decision Log** — use when: Every review someone disputes what was decided last time. Output: Per entry: the decision, date, decider, options considered, reason, what it changes, a review date and a superseded-by link. ### 6 · Run the delivery rhythm - **Project Management Plan** — use when: The process is heavier than the project, or there is none. Output: One page: the lifecycle picked with a reason, only the meetings and documents this project needs, and the reporting cadence and tolerances. - **Meeting Minutes** — use when: People remember decisions differently. Output: Decisions, actions with one owner and a due date, risks raised and open questions only, plus a chat-ready summary and a same-day send note. - **Sprint Planning** — use when: Someone else committed the team to more than fits. Output: One sprint goal written as an outcome, the sprint's capacity by role, the items the team selected with its own estimates, a definition of done check, and what was left out and why. - **Kanban Board** — use when: Everyone is busy and nothing finishes. Output: Columns that match the real workflow, WIP limits per column and blocked markers, the four flow measures computed, and the oldest items to swarm. - **Sprint Retrospective** — use when: The same retro actions repeat for months. Output: Last retro's action reviewed first, a format picked for this sprint's question, the team's input grouped into themes, one improvement with an owner and a check date, and the items that belong above the team. ### 7 · Report honestly - **Project Status Report** — use when: “On track” every week has stopped meaning anything. Output: A one-screen weekly report with a subject line that carries the news, the asks at the top, and a RAG rating backed by written definitions and evidence. - **Burndown Chart** — use when: Leaders do not trust the dashboard. Output: A burndown and a burnup from the tracker export with the scope line shown separately, a two-sentence reading of the trend, and what the chart cannot show. - **Earned Value Management** — use when: There is a budget and a baseline, and you need a number. Output: A planned value, earned value and actual cost table, SPI, CPI and a forecast at completion in plain words, and a one-line verdict for the status report. - **Communication Plan** — use when: Sixty people got the go-live email and few opened it. Output: An audience table, one cut of the status per audience, go-live and change announcements, and a check that the key messages were read. ### 8 · Close well - **Lessons Learned** — use when: The lessons register is full and nobody reads it. Output: A blameless review of what was planned against what happened and why, what to sustain and improve, and each lesson turned into an owned action or a template change. - **Project Closure Report** — use when: The team is moving on and the project never formally ends. Output: Objectives against results, acceptance status per deliverable, final schedule and cost against baseline, open items with new owners, a benefits owner with review dates, and sponsor sign-off. - **Project Handover** — use when: The project passes to operations or another project manager. Output: Acceptance criteria met or not, open items and known issues with workarounds, contacts by role, the support period, access and documents, and the receiver's confirmation checklist. ## Workflows - **Agree the scope** (A scope nobody agreed, and a "small" ask on its way.) Project Charter → Project Scope Statement → Change Request Form - **Estimate and schedule** (You need a schedule built on ranges.) Work Breakdown Structure → Three-Point Estimate → Critical Path → Gantt Chart - **Catch risks early** (While the risks are still cheap.) Pre-Mortem Analysis → Risk Register → RAID Log - **Report honestly** (A status report is due, and it has to mean something.) Burndown Chart → Earned Value Management → Project Status Report → Escalation Email → Steering Committee Deck - **Close well** (The project is ending and needs a proper close.) Lessons Learned → Project Closure Report → Project Handover ## The quality bar - One artifact per skill, the same shape every time, and it says what to paste, when not to use it and when it is done - The maths is done, not named: PERT ranges with spread, the critical path with float, SPI and CPI with a forecast - Every number comes with the case where it misleads, so nobody reports a figure they cannot defend - Every status rating cites its evidence in words, and nothing is marked green that the evidence does not support - Every decision, action and risk has a named owner and a date, and every log entry has a review date - Nothing grades, ranks or profiles people: capacity is shown by role and skill, stakeholders by decision rights, never by attitude or output per person - The red line: Claude plans and tracks — the people doing the work commit to the dates, and no status goes green to dodge a hard conversation ## Questions people ask ### What is in the AI for Project Management pack? 31 Claude skills that follow the life of a project in eight stages: start, stop scope creep, estimate and schedule, catch risks, get decisions made, run the delivery rhythm, report, and close. Each skill is one artifact you walk away with. ### How do I install these Claude skills? In Claude Code, two lines install all 31 at once. In Claude.ai or Claude Desktop, download the zip, open Settings, Skills, Add, and upload the skills you want. Either way it takes about five minutes. ### Do I need Claude Code to use them? No. The skills work in Claude.ai, Claude Desktop and Claude Code. In the app you upload them yourself; in Claude Code one command installs the whole pack. ### Do the skills work with ChatGPT or other assistants? They are built as Claude Skills, and installing them is a Claude feature. Each one is a plain Markdown file, so you can read it and reuse the method elsewhere, but it will not load and trigger by itself outside Claude. ### Will Claude commit to dates for the team? No. Claude plans and tracks. The people doing the work commit to the dates, and no status goes green to dodge a hard conversation. Each skill ends in the decision a named person makes, by a date. ### Is the pack free? Yes. It is free to use inside your company and in client work, and not for resale. There is no signup: the download and the install command are on this page. ## About Written by Polar Bear (https://meet-polar-bear.com/). Published on Lantern (https://lantern.meet-polar-bear.com/), which turns your notes, documents and call transcripts into social post drafts you review and publish.