--- name: Your Own Review category: career works_with: [claude, chatgpt, gemini] difficulty: plug-and-play version: 1.0 source: original (built for this library; guardrails per the TemplatedAI standard) tested: no --- # Your Own Review ## What it does Writes the self-assessment nobody can write about themselves — turning a year you half-remember into specific contributions, without either the bragging that makes you cringe or the modesty that erases you. Built only from things you actually did. ## When to use it - Review season, and the form is due Friday - You genuinely can't remember what you did in March - Your last self-review said "worked hard and supported the team" ## The skill ``` Act as an editor helping me write an honest self-assessment. HARD RULES: invent nothing — no metrics, no impact figures, no achievements I didn't describe; if a claim needs a number I don't have, tell me to go find it rather than estimating one. Don't inflate and don't let me erase myself — flag both. Keep my voice; this gets read by people who know how I talk. No corporate filler verbs where a plain one works. My role: [WHAT I DO] The period: [WHAT THIS REVIEW COVERS] What I did, in whatever order it comes out — dump everything: [PROJECTS, FIXES, THINGS THAT WENT WRONG, THINGS I LEARNED, PEOPLE I HELPED, ANYTHING] What I have evidence for: [NUMBERS, DOCS, FEEDBACK, SHIPPED THINGS] What my company says it values: [COMPETENCIES, GOALS, OR "no idea"] Build: 1. THE MEMORY PASS: questions to jog what I've forgotten — the quiet saves, the thing that didn't break because I fixed it, the person I unblocked. Under-recorded work is most work. 2. THE CONTRIBUTIONS: four to six items, each as what changed because I did it, not what I was assigned. Mark clearly which ones need a number I must go and find. 3. THE HONEST WEAKNESS: how to name a real development area without handing over ammunition — a genuine gap with what I'm doing about it, not a fake strength in a humble costume. 4. WHAT WENT WRONG: if I mentioned a failure, how to include it as something I learned from. A review with no failures in it reads as either dishonest or uninvolved. 5. THE MAPPING: line my contributions up against what my company says it values, using their words where they genuinely fit — and tell me where they don't rather than forcing it. 6. THE ASK: what I want next — scope, project, growth — stated once, clearly, at the end. Keep it to the length I'd actually read if I were my manager. ``` ## Example output [TO FILL AFTER TESTING] ## Tweaks - Keep a running note all year — three lines a month beats reconstructing twelve months in a panic on Thursday night - "What changed because I was there" is the only question that matters; everything else is job description - Name one real weakness. Reviewers discount a document with none, and you lose the credibility that carries the rest - Never invent a metric to sound stronger. One challenged number puts every other claim in doubt --- --- name: Get Real Feedback category: career works_with: [claude, chatgpt, gemini] difficulty: plug-and-play version: 1.0 source: original (built for this library; guardrails per the TemplatedAI standard) tested: no --- # Get Real Feedback ## What it does Gets you something more useful than "yeah, all good" — by asking questions specific enough that a conflict-avoidant manager can answer them honestly, and by making it safe to tell you the thing they've been not telling you. ## When to use it - Every one-to-one ends with "no concerns, keep it up" - You were passed over and nobody will say why - You want to improve and have no idea at what ## The skill ``` Act as a coach helping me extract useful feedback. HARD RULES: you have no knowledge of my performance or my manager — never guess what they think of me or what the "real reason" was; help me ask, not speculate. No manipulation and no traps designed to corner someone into criticising me. If I describe something that sounds like bullying or discrimination rather than awkward management, say so plainly and point me to the right channels instead of coaching me to try harder. Who I'd ask and what they're like: [MANAGER / PEER / SKIP-LEVEL — direct? avoidant? busy?] What I've asked before and what I got: [THE NON-ANSWERS] What I actually want to know: [PROMOTION? A SPECIFIC WORRY? GENERAL IMPROVEMENT?] What I suspect but can't confirm: [HONESTLY — held as a question] Give me: 1. WHY "ANY FEEDBACK?" FAILS: in one line, so I stop using it. 2. THE SPECIFIC QUESTIONS: five, each narrow enough to be answerable and awkward enough to be useful — about one project, one behaviour, one decision. Include the two that most reliably produce a real answer ("what would you have done differently in my position on X?" style) and say why they work. 3. THE PERMISSION LINE: how to open so an avoidant person believes I actually want it — and what I must not do when I hear it. 4. THE FUTURE FRAME: how to ask what would need to be true for the promotion, scope, or raise I want — turning a verdict on the past into a checkable list for the future. 5. RECEIVING IT: what to say in the moment, especially to something that stings. The one question that gets a vague criticism ("more executive presence") turned into an observable behaviour I could actually change. 6. THE FOLLOW-UP: what to send afterwards so what was said exists in writing, and how to come back in a month having done it. ``` ## Example output [TO FILL AFTER TESTING] ## Tweaks - "Any feedback for me?" is a yes/no question and busy people answer it with "no" - The most productive question in the set: "what would you have done differently in my position?" — it invites critique of a decision rather than of a person - Any feedback you can't observe isn't yet feedback. Ask what it would look like if you did it right, until you get something you could film - Never argue in the moment, even when it's wrong. Thank them, go away, decide later — argue once and you close the tap permanently --- --- name: Ask for a Raise category: career works_with: [claude, chatgpt, gemini] difficulty: plug-and-play version: 1.0 source: original (built for this library; guardrails per the TemplatedAI standard) tested: no --- # Ask for a Raise ## What it does Builds the case before the meeting: what you've actually delivered, the number you're asking for and where it came from, and what you'll say to each answer including no. One hard rule throughout — the AI never tells you what you're worth. Pay is local, current, and specific, and a confident invented figure is the fastest way to lose the room. ## When to use it - You've been meaning to ask for eleven months - Your job grew and your pay didn't - A review is coming and you don't want to improvise ## The skill ``` Act as a preparation partner for a pay conversation. HARD RULES: NEVER state a salary figure, market rate, or percentage as fact — your training data is stale and pay is local; instead tell me exactly where to find real current numbers for my role, country, and industry, and make me bring them. Never invent achievements, metrics, or figures for my case — if I haven't given you a number, don't produce one. Never state employment law or notice rights; name what to verify. No manipulation, no fake competing offers, no threats I don't intend to keep. My role and how long I've been in it: [DETAILS] What I actually deliver — with real numbers where I have them: [WHAT YOU DID, NOT YOUR JOB DESCRIPTION] What's changed since my pay was set: [SCOPE, HEADCOUNT, RESULTS] What I currently earn and what I'm thinking of asking: [OR "no idea"] Who decides, and what I know about the constraints: [BUDGET CYCLE, FREEZES, HOW THEY DECIDE] Help me: 1. THE EVIDENCE. Turn what I told you into three or four specific contributions — what changed because I was there, in numbers where I have them and plain facts where I don't. Flag anything vague and ask me for the specific version. 2. THE HOMEWORK I MUST DO. Where to find real pay data for MY role and location — the sites, the surveys, the people to ask, how to ask a recruiter without applying. Tell me plainly that any figure you produced yourself would be fiction. 3. THE NUMBER. Once I bring real data: how to choose what to ask for, why a specific figure beats a range, and how to say it without flinching or apologising. 4. THE ASK. How to open the meeting and the two sentences that make the request. Short. No preamble about how long I've been thinking about this. 5. THE ANSWERS. What to say to: yes; "not now, no budget"; "let's revisit in six months"; and a counter lower than I asked. For each, the specific follow-up question that turns a vague deferral into something checkable — a date, a number, a condition. 6. AFTERWARDS. What to put in writing the same day, whatever the answer, so this conversation exists next time. ``` ## Example output [TO FILL AFTER TESTING] ## Tweaks - Bring real market data or don't mention market at all. A number you can't source gets dismantled in one question - Ask for a specific figure. A range is an invitation to hear the bottom of it - "Not now" is only useful once it has a date and a condition attached — get both before you leave the room - Email a short summary afterwards. Six months from now, the memory of the meeting will otherwise belong to whoever benefits from forgetting it --- --- name: Decline Extra Work category: career works_with: [claude, chatgpt, gemini] difficulty: plug-and-play version: 1.0 source: original (built for this library; guardrails per the TemplatedAI standard) tested: no --- # Decline Extra Work ## What it does Turns "I'm already drowning" into a professional answer that protects your workload without reading as a refusal — by making the trade-off visible. Most extra work is assigned by people who don't know what's already on your plate. ## When to use it - A new "quick thing" arrives on top of a full week - You're the person who always says yes and it's stopped being a compliment - You said yes last time and something else quietly slipped ## The skill ``` Act as a level-headed colleague helping me respond. HARD RULES: never invent workload, deadlines, or reasons — work only from what I list; don't help me dodge work that's genuinely my job, and say so if that's what this is; no passive-aggression and no martyrdom. You don't know my workplace's politics — offer the professional version and let me judge the room. If I'm describing a pattern of being overloaded to the point of harm, name it rather than optimising another polite decline. What's being asked: [THE NEW TASK, AND BY WHOM] What's already on me, with deadlines: [LIST IT] Who decides my priorities: [SAME PERSON? SOMEONE ELSE? NOBODY?] Is this actually my job: [YES / NO / GREY AREA] What happens if I just say yes: [WHAT SLIPS] Give me: 1. THE HONEST READ: from what I listed, is this reasonable work I should absorb, a priority question for someone else to answer, or genuinely not mine? Say which. Don't help me decline something I ought to do. 2. THE TRADE-OFF REPLY: the response that doesn't say no but makes the cost visible — what's on my list, what would move, and asking them to choose. This is usually the right answer, because they often don't know. 3. THE CLEAN NO: for when it isn't mine — short, professional, no apology spiral, with the right person or route named if I know it. 4. THE YES WITH TERMS: if I'm taking it, what to confirm in the same message — what drops, what date moves, what help I need. Unconditional yeses are how workloads quietly compound. 5. WHAT TO PUT IN WRITING: the one-line summary that makes the agreed trade-off exist later, when someone asks why the other thing was late. 6. THE PATTERN CHECK: if this keeps happening, what to raise separately — as a workload conversation, not as a reaction to one more task. ``` ## Example output [TO FILL AFTER TESTING] ## Tweaks - "Happy to take it — which of these three should slip?" is the whole skill. It's a yes, and it hands the priority decision to the person whose decision it is - Most overload is invisible to the person assigning it; showing the list is informative, not defensive - Confirm the trade-off in writing the same day. Verbal agreements evaporate exactly when you need them - If every week needs this skill, the problem isn't the tasks — raise it as its own conversation, using Get Real Feedback or Set a Boundary --- --- name: Raise a Problem category: career works_with: [claude, chatgpt, gemini] difficulty: plug-and-play version: 1.0 source: original (built for this library; guardrails per the TemplatedAI standard) tested: no --- # Raise a Problem ## What it does Helps you say the thing upward — the plan that won't work, the deadline that's fiction, the decision heading somewhere bad — so it gets heard as useful rather than as complaining or insubordination. Disagreement is only valuable if it arrives in a form someone can act on. ## When to use it - You can see it coming and everyone's nodding - You flagged it once, vaguely, and nothing happened - You're writing an angry message you shouldn't send ## The skill ``` Act as a calm advisor helping me raise a concern upward. HARD RULES: you don't know my organisation, my boss, or the politics — never guess motives, and present options rather than telling me what will happen; never invent evidence, and if my case rests on a feeling rather than something observable, say so and help me find the observable version. No sarcasm, no score-settling. If what I'm describing is safety, fraud, harassment, or anything unlawful, say so plainly: that isn't a disagreement to finesse, it goes through the proper channel, and tell me to find out what that channel is where I work. What I think is wrong: [THE PROBLEM] What I've actually observed: [FACTS, NOT INFERENCES] Who decided it and what I know about why: [CONTEXT] What happens if nobody changes course: [THE CONSEQUENCE] What I've said so far: [AND HOW IT LANDED] How much risk I can carry here: [NEW JOB? SECURE? ALREADY LEAVING?] Help me: 1. SEPARATE FACT FROM READING: sort my account into what I observed and what I concluded. Only the first half travels. 2. THE ONE-SENTENCE PROBLEM: what's wrong and what it will cost, in a sentence a busy person can repeat to someone else. If I can't get it into a sentence, I'm not ready to raise it. 3. THE ALTERNATIVE: what I'd do instead — because a problem without a proposal is a complaint, and gets filed as one. If I don't have one, help me say that honestly rather than pretending. 4. THE ROUTE: privately first or in the room, to my manager or further, in writing or spoken — with the trade-off of each, given the risk level I stated. 5. THE OPENING: how to start so it lands as help. Two versions: one that assumes good faith and offers information, one firmer for when I've already raised it and nothing happened. 6. WHEN THEY GO AHEAD ANYWAY: how to disagree and commit without either sulking or pretending I agreed — and what one line to put in writing so the record is accurate without being a trap. ``` ## Example output [TO FILL AFTER TESTING] ## Tweaks - Lead with the cost, not the flaw. "This will miss the March date" gets attention that "the plan is wrong" does not - Bring an alternative, even a weak one. It converts you from critic to participant in a single sentence - One clear written note, once, is not politics — it's how organisations avoid repeating things - Disagree and commit is a real move: say your piece properly, then genuinely help it work. Dragging your feet afterwards costs you the right to be listened to next time --- --- name: Quit Well category: career works_with: [claude, chatgpt, gemini] difficulty: plug-and-play version: 1.0 source: original (built for this library; guardrails per the TemplatedAI standard) tested: no --- # Quit Well ## What it does Gets you out the door with the reference, the relationships, and your own self-respect intact: the resignation message, the notice-period plan, what to say in the exit interview, and the counter-offer question answered before it's asked. ## When to use it - You've decided, and now you have to tell someone - A counter-offer is coming and you want to know your answer in advance - You're leaving angry and would like to not act on that ## The skill ``` Act as a steady advisor for resigning professionally. HARD RULES: never state employment law, notice periods, holiday pay, or contractual rights as fact — these vary by country and contract; tell me exactly what to check in my contract and who to ask. Never draft anything that burns a bridge, however justified I feel — and if I want to vent, say plainly that the resignation letter is the wrong place. Don't invent reasons for me. If I describe something unlawful or a serious grievance, say that it needs a proper channel and, where money or rights are at stake, a professional — not a well-worded goodbye. Where I work and my role: [DETAILS] Why I'm leaving, honestly: [FOR MY CLARITY, NOT FOR THE LETTER] What I want afterwards: [REFERENCE? RELATIONSHIPS? NEVER AGAIN?] What's signed and what's next: [OFFER ACCEPTED? START DATE?] How I expect them to react: [FINE / BADLY / COUNTER-OFFER] Help me: 1. THE CHECK FIRST: what to read in my own contract before I say anything — notice, holiday, bonus timing, anything repayable, restrictions on where I go next. Tell me to verify each with HR or a professional rather than trusting any general answer. 2. THE ORDER: who hears it first, and why telling a colleague before my manager rarely ends well. 3. THE MESSAGE: short, warm, factual. Leaving, last day, thank you, offer to hand over. No grievances, no explanation of why — that belongs in a conversation, if anywhere. 4. THE CONVERSATION: what to say out loud, including how to answer "why?" honestly without turning it into a list of complaints I can't take back. 5. THE COUNTER-OFFER: decide my answer NOW, before it's offered — the questions to ask myself about what would actually have to change, and why money alone rarely fixes the reason people leave. 6. THE LAST WEEKS: a handover that means my name is remembered well — documentation, the list of things only I know, who needs briefing. This is what a reference is actually made of. 7. THE EXIT INTERVIEW: what's worth saying, what isn't, and being clear-eyed that it's rarely confidential. ``` ## Example output [TO FILL AFTER TESTING] ## Tweaks - Nothing goes in writing that you wouldn't want read aloud by a stranger in two years — because it might be - Decide the counter-offer answer before you resign. In the room, flattered and relieved, is the worst moment to weigh it - The handover is the reference. Weeks of visible, generous handover outlast whatever anyone thought of your last project - Keep the real reasons for a conversation with people you trust. The exit interview is a company record, not a confessional --- --- name: The Work Year category: career tools_required: [Any AI chat, and one running note you keep all year] cost: free time_to_set_up: three minutes a month, plus an hour before each review version: 1.0 source: original (built for this library; guardrails per the TemplatedAI standard) tested: no --- # The Work Year ## What it does Chains six skills into the loop that actually moves a career: record what you do, review it honestly, find out how you're really seen, ask for what that earns, protect your workload while you do it — and, when the answer stays no, leave properly. The whole cycle runs on evidence you gathered before you needed it. ## What you need - One running note — phone, doc, anywhere — where three lines a month live - The six skills: Your Own Review, Get Real Feedback, Ask for a Raise, Decline Extra Work, Raise a Problem, Quit Well ## Setup (one time) 1. **Start the note today.** Date, what changed because you were there, any number attached. Three lines. This one habit is the difference between a self-review written from evidence and one written from panic on a Thursday night. 2. **Find out when decisions actually happen** — the budget cycle, the promotion round. Asking two weeks after the money is allocated is the most common reason a good case fails. ## The workflow (per year) 3. **Monthly:** three lines in the note. That's the whole maintenance cost. 4. **Quarterly — Get Real Feedback.** Specific questions, one project at a time. A surprise in a formal review means this step was skipped. 5. **Before the review — Your Own Review.** Build it from the note. Name one real weakness; a review with none reads as unserious. 6. **Ahead of the budget cycle — Ask for a Raise.** Bring real market data you sourced yourself. Whatever the answer, write it down the same day; "not now" only counts once it has a date and a condition. 7. **All year — Decline Extra Work** when the plate is full, and **Raise a Problem** when you can see something coming. Both are how you stay effective enough to have a case at all. 8. **If nothing moves after a full cycle of this — Quit Well.** Not in anger, and not without the handover that becomes your reference. ## Maintenance habits - Put every outcome in writing the same day. Memory of who promised what belongs to whoever benefits from misremembering - Never bring a pay number you can't source; one challenged figure discredits the whole case - Don't save up problems for review season — a year-old concern raised late lands as a grudge - Leave on good terms even when they don't deserve it. The industry is smaller than it looks and references outlive grievances ## What we stripped This pack is original, but the career-advice genre it sits beside is built on things we refuse to publish. No salary figures or "market rates" from an AI — pay is local and current, our training data is neither, and a confident wrong number loses you the room; every relevant skill sends you to real sources instead. No invented achievements or metrics to strengthen a case. No scripts for manipulating a manager, no fake competing offers, no leverage plays. No statements about employment law, notice, or your rights — those vary by country and contract, so we tell you what to check and who to ask. And no promises: a well-built case can still get a no, because budgets and decisions aren't yours to control. What the loop guarantees is that you'll know where you stand and have the evidence to act on it.