These are real open-source skills built in the open, laid out like a small PMO so you can see who does what. Each entry tells you plainly what the skill does and when to grab it, credits the maker, and shows you how to paste the file into a Claude Project. The one thing we wrote ourselves sits on the status desk: a brief that keeps RAG honest and will not invent a percent complete, a date, or a Green over a Red blocker.
How to install any skill on this page
This set is paste-first. We do not invent one-line installs that are not there.
- The paste route (use this). Open the skill's SKILL.md file in its GitHub repo, copy the whole file, and paste it into a Claude Project. Same instructions, no command line. We link every file below.
- Optional plugin route. The pm-claude-skills repo is plugin-structured, so if you already use Claude Code you can add it as a plugin from its GitHub repo. Paste is the reliable method on this page, since two of the skills below sit outside the plugin folder.
- Already have the freelance set? That page has a lighter weekly report and SOP. This page is the PM version: RAG, owners, scorecards. Cross-link, do not paste both for the same job.
1. Status
The desk that tells the truth in one page. Starts with our own brief, then one open-source RAG skill.
The tidy-status brief by AI with Adele
The one thing on this page we built ourselves. It writes a weekly status from facts you actually have: a real RAG, named blockers, and decisions someone can take. Save it as tidy-status.md, paste it into a Claude Project, and it will refuse invented percents, fake dates, and a Green rating sitting on top of a Red blocker.
--- name: tidy-status description: Writes an honest weekly project status. Asks for real RAG, blockers, and decisions before drafting. Refuses invented percent-complete, fake dates, and a Green rating over a Red blocker. US English. A skill drafts, you decide. --- # The tidy-status brief You are my status editor. Your job is one page a busy sponsor can read in a minute: specific, honest, and short. What is at stake is trust. A status that invents 85% done, a recovery date, or a Green over a Red blocker will get me a surprise later. You never invent a percent, a date, a budget figure, or a RAG colour I did not give. FIRST, CHECK. Ask me these and wait. Do not draft until I have answered: - What is the project called, and who is this for (sponsor, team, steering)? - What is this period (dates I actually have)? - Overall RAG: Red, Amber or Green? I must pick. If I hedge, help me pick from the facts, do not pick for me. - What shipped this period? What was due and did not? - What is blocked, who owns it, and what decision is needed? - Any budget number I am willing to publish? If not, write [gap: budget]. Never invent a variance. - Next period: only the 3-5 things I named. THE REFRAME, teach me this in plain words before the draft: Most AI status reports fail because they sound busy and they hide the Red. A good status is a decision tool: here is the colour, here is why, here is who must do what by when. Milestones are complete or not. If I cannot name a date, we write [TBD]. THE METHOD: 1. Lead with the RAG and the one-line why. If it is Red, say Red in the first sentence. 2. Milestones are binary: complete or not. Never write 85% done. 3. Every blocker has a named owner or is marked unassigned. Never invent a person. 4. Dates: only ones I gave. Otherwise [TBD]. 5. Decisions needed: who must decide, by when, and my recommendation only if I gave one. 6. Keep it one page. No theatre. OUTPUT, in this order: - One-line RAG and why. - The draft status. - Every [gap] or [TBD] I still owe. - The sentence most likely to hide bad news. Delete it if I am even slightly unsure. WORKED EXAMPLE (illustrative, made up: do not copy the facts onto my status): Project: new booking form. Period: this week. RAG: Amber because the payment test is blocked on a vendor reply. Shipped: copy. Not shipped: live test. Owner: Sam. Decision: go/no-go Friday or [TBD]. Draft refuses "85% complete" and invents no go-live date. RULES: NEVER invent a percent, a date, a budget, a RAG, or an owner. If a claim needs a number I have not given, write [gap] or [TBD]. US English, plain words. No em dashes in the draft unless my sample uses them. A skill drafts, I decide. SELF-CHECK: end by naming the sentence most likely to soften a Red, the date you are least sure is true, and whether RAG is mine or still [gap].
project-status-report by mohitagw15856
A full RAG pack: overall / schedule / budget / scope / risk, binary milestones, blockers with owners, and decisions required. Explicitly refuses "85% done" and will not bury a Red after good news. Grab it when the audience is a steering group rather than a Friday Slack. MIT. Uses placeholder amounts; you fill the numbers.
Paste: project-status-report/SKILL.md.
2. Meetings
The desk that keeps the commitment from evaporating after the call.
meeting-notes-to-actions by Skill Me
Four-part record from a paste: summary, decisions, action table, open questions. Ownerless items stay UNASSIGNED. Unstated dates stay TBD. Will not invent a task. Grab it when you need a recap plus the table. MIT.
meeting-action-extractor by mohitagw15856
Stricter on commitments versus wishes. Flags "someone should" as an orphan instead of turning chatter into tickets. Will not invent an owner or a due date. Grab it when the notes are a transcript and you only want the actions. MIT.
3. Plans and process
The desk that writes down how the work actually runs, with named owners.
process-documentation by mohitagw15856
A process someone new can follow: roles, numbered steps, tools, edge cases, common mistakes, and an escalation path. Every step needs a named role, not "the team". Grab it when the work has handoffs. MIT. For a shorter personal SOP, the freelance set already has nawazxz's sop-writer.
Paste: process-documentation/SKILL.md.
4. Vendors
The desk that compares two quotes without falling in love with the demo.
vendor-evaluation by mohitagw15856
Weighted scorecard (must total 100%), rubric, questions, reference checks, then a recommendation with risks and conditions. You supply the weights and the scores. Grab it before you sign, not after the kickoff. MIT.
Paste: vendor-evaluation/SKILL.md.
Where the line is
- These are other people's work, credited and open. Third-party skills here are MIT-licensed. Named makers only. Study them, paste them, credit them. Do not lift their text into your own page and call it yours.
- A skill drafts, you decide. Read every line before a status sends or a scorecard is shared. Never publish a first draft unedited.
- Never invent a percent, a date, a RAG, or an owner. Our status brief leaves [TBD] rather than make a recovery date up. The meeting skills here flag UNASSIGNED instead of guessing who promised.
Pick a desk and paste one today
Do not fit out the whole PMO at once. If Friday needs a status, start with our brief. If the notes are a pile, try the Skill Me extractor. If two vendors are in a bake-off, use the vendor-evaluation scorecard. Recaps that still sound like a brand account need a voice sample first: make AI write in your voice.
Looking for skills for a different job? Head back to the hub: AI skills by role. Marketing, sales, HR, support and freelance are live too: AI marketing skills, AI sales skills, AI HR skills, AI support skills, AI freelance skills.
A few quick questions
Are these project and ops skills free?
Yes. The third-party skills on this page are open-source, MIT-licensed and free to use. Our own tidy-status brief is free to copy on this page.
Do I need to use the terminal?
No. Every skill on this page is paste-first. Open the skill's SKILL.md on GitHub, copy the whole file, and paste it into a Claude Project. This library can also install as a Claude Code plugin if you already use that route.
What is the tidy-status brief on this page for?
It is our own brief for a weekly status that stays honest. It asks for the real RAG, the blockers, and the decisions needed. It will not invent a percent complete, a date, or a Green rating over a Red blocker.