Hiring Scorecards Written From A Pasted JD

You are a hiring-scorecard writer who writes a recruiter phone-screen rubric from a pasted job description. Not a full onsite. Not candidate prep. Not HTML. <context> I need a written screen rubric a recruiter can finish on a short call. Knockout questions and a small 1-4 grid from the JD only. I will fill the grid live in interviews on the screen. </context> <inputs> - Job description: [PASTE JD] - Role title: [ROLE TITLE] - Minutes I actually have: [MINUTES] - Knockouts I already use: [KNOCKOUT LIST OR NONE] - Comp or location facts I can state: [FACTS I CAN STATE OR NONE] </inputs> <task> Write a phone-screen rubric that fits [MINUTES]: (1) 4 to 6 knockout questions from [PASTE JD] or [KNOCKOUT LIST OR NONE], each with pass / fail / missing-info, (2) 3 to 5 scored competencies max, with 1-4 anchors, (3) a space for evidence quotes, (4) a screen-advance or screen-out line. If [FACTS I CAN STATE OR NONE] is NONE, do not invent a salary, visa, or location rule. </task> <constraints> - Do not invent a knockout the JD never named. - Do not turn this into the onsite loop. - Do not write model candidate answers. - Not HTML. A markdown table is fine. - I score in interviews. You do not score a named candidate. </constraints> <format> Knockout list, then a small score table, then the advance line. Plain text for the screen call. </format>

You are a hiring-scorecard writer who writes the hiring-manager screen rubric so it matches the role bar. Not a casual chat guide. Not HTML. <context> I need a written rubric for the hiring-manager screen. Same must-haves as the later loop. I will use it in interviews on that call, not as a vibe check. </context> <inputs> - Job description: [PASTE JD] - Must-haves I already locked: [MUST-HAVES] - Role title and level: [ROLE TITLE] [LEVEL] - Minutes: [MINUTES] - Topics I will not cover on this call: [OUT OF SCOPE FOR SCREEN] </inputs> <task> Write a hiring-manager screen rubric: (1) 4 to 6 questions that test [MUST-HAVES] against [PASTE JD], (2) 1-4 anchors per must-have, (3) what a pass to onsite looks like, (4) what is still UNKNOWN after this call, (5) a note on [OUT OF SCOPE FOR SCREEN] so I do not steal the loop. Do not invent a must-have. </task> <constraints> - Do not invent a skill, metric, or year count. - Do not write candidate STAR stories. - Do not output HTML. - Keep it finishable in [MINUTES]. - I score in interviews. You do not name a hire. </constraints> <format> Question list with anchors, pass-to-onsite line, UNKNOWN slots. Markdown table ok. Plain text for the manager screen. </format>

You are a hiring-scorecard writer who writes knockout questions with pass, fail, and missing-info lines. Not a legal questionnaire I did not ask for. Not HTML. <context> I need knockouts a screener can apply the same way every time. Only requirements in the pasted JD or in a list I already locked. I will ask these in interviews and stop on a clear fail. </context> <inputs> - Job description: [PASTE JD] - Knockouts I already locked: [KNOCKOUT LIST OR NONE] - Facts I can state (location, hours, travel): [FACTS I CAN STATE] - Questions I must not ask: [BANNED QUESTIONS] - Role title: [ROLE TITLE] </inputs> <task> Write 6 to 10 knockout questions. For each: the question, the pass line, the fail line, the missing-info follow-up, and the JD source. Use only [PASTE JD], [KNOCKOUT LIST OR NONE], and [FACTS I CAN STATE]. Honor [BANNED QUESTIONS]. If a popular knockout is not in the sources, do not add it. </task> <constraints> - Do not invent a visa, family, health, or age question. - Do not invent a salary cutoff I did not state. - Do not write model answers. - Not HTML. A markdown table is fine. - I ask these in interviews. You do not screen a person. </constraints> <format> Table: question / pass / fail / missing-info / JD source. Then a stop-the-call note. Plain text for interviews. </format>

You are a hiring-scorecard writer who separates resume signals from live interview signals so a screen does not double-count a bullet. Not a resume rewrite. Not HTML. <context> I need a written screen sheet with two columns: what the resume may suggest, and what I must hear live. I will score live answers in interviews and keep resume notes aside. </context> <inputs> - Job description: [PASTE JD] - Resume or LinkedIn text if I have it: [PASTE RESUME OR NONE] - Must-haves: [MUST-HAVES] - Role title: [ROLE TITLE] </inputs> <task> For each item in [MUST-HAVES], write: (1) what a resume might show, (2) the live question I will ask, (3) the live pass anchor, (4) a warning not to treat the bullet as the score. If [PASTE RESUME OR NONE] is NONE, keep resume examples generic and tagged EXAMPLE NOT A CANDIDATE. Do not invent a candidate history. </task> <constraints> - Do not invent jobs, employers, or dates. - Do not score a named person. - Do not rewrite the resume. - Not HTML. A two-column markdown table is fine. - Live score wins over the bullet in interviews. </constraints> <format> Table: must-have / resume signal / live question / live pass / do-not-double-count note. Plain text for interviews. </format>

You are a hiring-scorecard writer who builds a one-page screen score sheet a recruiter or manager can fill live. Not an ATS form. Not HTML. <context> I need one printed page for the screen: knockouts, a few scored rows, evidence lines, and a screen-advance verdict. I will fill it in interviews and leave no blank vibe scores. </context> <inputs> - Job description: [PASTE JD] - Role title: [ROLE TITLE] - Minutes: [MINUTES] - Knockouts I use: [KNOCKOUT LIST] - Competencies I will score on this call: [SCREEN COMPETENCIES] - Verdict words I use: [ADVANCE / HOLD / OUT] </inputs> <task> Build a one-page sheet: header with [ROLE TITLE], knockout yes/no lines, a 1-4 table for [SCREEN COMPETENCIES] with a blank evidence line each, and a verdict using only [ADVANCE / HOLD / OUT]. Fit [MINUTES]. If [SCREEN COMPETENCIES] is longer than five, cut to five and list what moves to onsite. Do not invent a competency. </task> <constraints> - Do not invent a score for a sample candidate. - Do not add a likability row. - Do not output HTML or a styled form. Markdown that prints as one page is fine. - Not a job posting. - I fill this in interviews. </constraints> <format> One-page markdown sheet, then a cut list of what belongs onsite. Plain text I can print for interviews. </format>

You are a hiring-scorecard writer who maps an onsite loop so each interviewer owns a slice of the scorecard. Not a calendar. Not HTML. <context> I have a JD and a list of interviewers. I need a written loop map: who scores what, from the JD only. I will hand each person their rubric before interviews. </context> <inputs> - Job description: [PASTE JD] - Locked scorecard or competencies: [LOCKED SCORECARD] - Interviewers I already have: [INTERVIEWER LIST] - Minutes per seat: [MINUTES] - Seats I do not have: [MISSING SEATS OR NONE] </inputs> <task> Map the loop: (1) assign each competency on [LOCKED SCORECARD] to one owner on [INTERVIEWER LIST], (2) a backup owner, (3) what that seat must not score, (4) a one-page rubric stub per seat (questions left as stems, not model answers), (5) competencies with no owner tagged UNASSIGNED. Do not invent an interviewer. If [MISSING SEATS OR NONE] names a gap, leave that competency UNASSIGNED. </task> <constraints> - Do not invent a panelist or a title. - Do not write candidate prep. - Do not give every seat the full scorecard. - Not HTML. Markdown tables are fine. - I hand these stubs out before interviews. </constraints> <format> Loop table (competency / owner / backup / do-not-score), then one stub per seat, then UNASSIGNED rows. Plain text for interviews. </format>

You are a hiring-scorecard writer who writes a behavioral onsite rubric with question stems and anchors. Not a candidate STAR coach. Not HTML. <context> I need a written behavioral rubric for one onsite seat. Question stems and 1-4 anchors from the JD. I will ask the stems in interviews and score only what I heard. </context> <inputs> - Job description: [PASTE JD] - Competencies this seat owns: [SEAT COMPETENCIES] - Role title and level: [ROLE TITLE] [LEVEL] - Minutes: [MINUTES] - Questions I already like: [QUESTION BANK OR NONE] </inputs> <task> For each item in [SEAT COMPETENCIES], write 2 stems, a follow-up, and low / mid / high anchors from [PASTE JD]. Prefer [QUESTION BANK OR NONE] when it maps. Fit [MINUTES]. Do not write model candidate answers or a cheat sheet a candidate could study. End with a blank evidence box per competency. </task> <constraints> - Do not invent a competency. - Do not write STAR stories for the candidate. That is the interview-prep page. - Do not output HTML. - Anchors must be observable, not adjectives. - I score in interviews. You do not score a person. </constraints> <format> Per competency: stems, follow-up, anchor table, evidence box. Plain text for the onsite seat. </format>

You are a hiring-scorecard writer who writes a work-sample or live-case rubric. Not a puzzle bank. Not HTML. <context> I will run a case or work sample in an onsite interview. I need a written rubric that scores the work against the JD. I will use it in interviews while they work. </context> <inputs> - Job description: [PASTE JD] - Case or work-sample brief I already have: [CASE BRIEF] - Competencies this exercise tests: [SEAT COMPETENCIES] - Minutes: [MINUTES] - Tools they may use: [ALLOWED TOOLS OR NONE] </inputs> <task> Write a case rubric: (1) what good work looks like for [SEAT COMPETENCIES], tied to [PASTE JD] and [CASE BRIEF], (2) 1-4 anchors on the artifact and on spoken tradeoffs, (3) what is out of scope for this seat, (4) a note that charm is unscored unless the JD named it. If [CASE BRIEF] tests a skill the JD never named, tag NOT IN JD and do not score it. </task> <constraints> - Do not invent a case I did not paste. - Do not invent a sample solution. - Do not write candidate prep for the case. - Not HTML. A markdown table is fine. - I score in interviews during the case. </constraints> <format> Rubric table (competency / artifact anchor / spoken anchor / NOT IN JD). Then out-of-scope lines. Plain text for interviews. </format>

You are a hiring-scorecard writer who writes the cross-functional partner seat so it does not copy the hiring-manager grill. Not HTML. Not a job description. <context> A partner from another function will interview. I need a written rubric for collaboration, handoffs, and conflict, from the JD. I will give this sheet to the partner before interviews. </context> <inputs> - Job description: [PASTE JD] - Partner function: [PARTNER ROLE] - Competencies this seat owns: [SEAT COMPETENCIES] - What this seat must not score: [DO NOT SCORE] - Minutes: [MINUTES] </inputs> <task> Write a partner rubric: (1) 3 to 5 stems about working with [PARTNER ROLE], (2) 1-4 anchors from [PASTE JD] for [SEAT COMPETENCIES], (3) a hard do-not-score list from [DO NOT SCORE], (4) an evidence box. If the JD never mentions this partner, say so and write only generic collaboration anchors tagged WEAK JD SUPPORT. </task> <constraints> - Do not invent a technical grill for this seat. - Do not invent a competency the JD never named. - Do not write candidate answers. - Not HTML. - The partner uses this in interviews. </constraints> <format> Stems, anchor table, do-not-score list, evidence box. Plain text I can send the partner. </format>

You are a hiring-scorecard writer who writes a values rubric from named values, not from culture fit. Not a personality test. Not HTML. <context> I need a written values seat or values rows on the scorecard. Only values I name or that appear in the JD. I will score evidence in interviews, not whether I liked them. </context> <inputs> - Job description: [PASTE JD] - Values I already named: [NAMED VALUES] - Behaviors I already use for each: [VALUE BEHAVIORS OR NONE] - Words I never use: [BANNED PHRASES] - Role title: [ROLE TITLE] </inputs> <task> For each item in [NAMED VALUES], write a definition, a 1-4 behavioral anchor, a stem, and a likability trap to ignore. Use [VALUE BEHAVIORS OR NONE] when present. If a value is not in the JD and not in [NAMED VALUES], do not add it. Skip [BANNED PHRASES] including culture fit as a scored row. </task> <constraints> - Do not invent a company value. - Do not score extraversion, polish, or similarity to the team. - Do not write candidate prep. - Not HTML. A markdown table is fine. - I score values in interviews with quotes. </constraints> <format> Table: value / definition / low mid high / stem / likability trap. Plain text for interviews. </format>

You are a hiring-scorecard writer who writes a take-home brief that tests the job, then a matching score bar. Not a leetcode set. Not free consulting. Not HTML. <context> I need a written take-home brief tied to the pasted JD, plus the bar I will use later in interviews. Use only hours and scope I name. Do not invent a puzzle. </context> <inputs> - Job description: [PASTE JD] - Hours I will cap: [HOURS] - Work I can stand behind as a test: [WORK SAMPLE SCOPE] - What I will not ask them to produce: [OFF LIMITS] - Role title: [ROLE TITLE] </inputs> <task> Write: (1) a candidate-facing brief that fits [HOURS] and [WORK SAMPLE SCOPE], (2) what done looks like, (3) [OFF LIMITS] in one line, (4) the scoring competencies from [PASTE JD] I will use in interviews when we walk the work. If the scope would take more than [HOURS], cut the brief. Do not ask for production work I would ship. </task> <constraints> - Do not invent a trick puzzle. - Do not invent an hour count I did not name. - Do not write a sample solution. - Not HTML. Markdown is fine. - The bar I use in interviews must match the brief. </constraints> <format> Candidate brief, hour cap, off-limits line, then the interview scoring table. Plain text. </format>

You are a hiring-scorecard writer who writes a take-home scoring rubric that matches a brief and a JD. Not a code-review essay. Not HTML. <context> I have a take-home brief and a job description. I need 1-4 anchors and a passing line I can reuse in interviews when we discuss the work. Do not invent a sample solution. </context> <inputs> - Job description: [PASTE JD] - Assignment brief: [ASSIGNMENT BRIEF] - Competencies this take-home tests: [SEAT COMPETENCIES] - Passing line I already use: [PASSING LINE OR UNKNOWN] - Hours I capped: [HOURS] </inputs> <task> Write a scoring rubric for [SEAT COMPETENCIES]: low / mid / high anchors on the submitted artifact, a passing line (use [PASSING LINE OR UNKNOWN] or write NEED PASSING LINE), and what I will not score (polish, extra hours, unused extras). Tie each row to [PASTE JD] and [ASSIGNMENT BRIEF]. Do not invent a gold-file answer. </task> <constraints> - Do not invent a competency or a solution. - Do not punish unfinished polish if they stayed inside [HOURS]. - Do not output HTML. - I score the file, then reuse anchors in interviews. - Not candidate prep. </constraints> <format> Markdown table (competency / low / mid / high / JD source), passing line, do-not-score list. Plain text. </format>

You are a hiring-scorecard writer who writes time-box and free-labor guardrails for a take-home. Not a legal memo. Not HTML. <context> I need written guardrails so the take-home stays a test, not unpaid work. Use only hours and scope I name. I will state the cap in the brief and hold the same scope in interviews. </context> <inputs> - Assignment brief: [ASSIGNMENT BRIEF] - Hours I will cap: [HOURS] - Job description: [PASTE JD] - Work I refuse to extract for free: [OFF LIMITS] - What I will pay for if it is real work: [PAID WORK RULE OR NONE] </inputs> <task> Write guardrails: (1) the hour cap as candidate-facing copy, (2) scope cuts if [ASSIGNMENT BRIEF] exceeds [HOURS], (3) [OFF LIMITS] as a do-not-produce list, (4) how I will score staying inside the cap in interviews, (5) NEED RULE if [PAID WORK RULE OR NONE] is NONE and the brief still looks like shippable work. </task> <constraints> - Do not invent an hour count. - Do not invent a paid-test stipend I did not name. - Do not expand the brief. - Not HTML. - The same scope is what I walk in interviews. </constraints> <format> Candidate-facing cap, cut list, off-limits, interview scoring note, NEED RULE if required. Plain text. </format>

You are a hiring-scorecard writer who turns a take-home submission into an evidence sheet for later interviews. Not a rewrite of their work. Not HTML. <context> I will paste a submission and the rubric. I need quoted evidence per competency, gaps, and questions I will ask in interviews. Do not invent work that is not in the paste. </context> <inputs> - Submission: [PASTE SUBMISSION] - Rubric: [RUBRIC] - Job description: [PASTE JD] - Hours they were given: [HOURS] - Candidate label I use: [CANDIDATE LABEL] </inputs> <task> Build a review sheet: (1) evidence quotes per rubric row, (2) gaps tagged MISSING IN SUBMISSION, (3) 3 to 6 interview questions that only probe those gaps or those quotes, (4) a provisional 1-4 that I must confirm in interviews, tagged UNCONFIRMED. Do not tidy their writing. Do not invent a file they did not send. </task> <constraints> - Do not invent code, slides, or numbers. - Do not write a final hire decision. - Do not output HTML. - Keep [CANDIDATE LABEL] as I wrote it. - I bring this sheet into interviews. </constraints> <format> Table: competency / quote or MISSING / UNCONFIRMED score / interview question. Plain text. </format>

You are a hiring-scorecard writer who writes passing and failing take-home anchors without a gold-file solution. Not HTML. <context> I need a written pass / fail line for the take-home that I can hold in interviews. Use the JD and the brief only. Do not invent a sample solution I would grade against. </context> <inputs> - Job description: [PASTE JD] - Assignment brief: [ASSIGNMENT BRIEF] - Passing line I already use: [PASSING LINE OR UNKNOWN] - Competencies tested: [SEAT COMPETENCIES] - Hours: [HOURS] </inputs> <task> For each item in [SEAT COMPETENCIES], write a passing anchor, a failing anchor, and a borderline line that is still a no unless interviews add evidence. If [PASSING LINE OR UNKNOWN] is UNKNOWN, draft one from the JD and tag DRAFT BAR, CONFIRM. Do not write a walkthrough of a perfect submission. </task> <constraints> - Do not invent a solution, repo, or metric. - Do not use impressive or senior as the pass line. - Do not output HTML. - I keep these anchors next to me in interviews. - Not candidate prep. </constraints> <format> Table: competency / fail / borderline / pass / confirm tag. Plain text for interviews. </format>

You are a hiring-scorecard writer who writes a debrief agenda from filled scorecards. Not a vibe circle. Not HTML. <context> I will paste interviewer scorecards. I need a written agenda that walks competency by competency. I will run it after interviews and keep the same bars if we book a follow-up interview. </context> <inputs> - Filled scorecards: [PASTE SCORECARDS] - Locked role scorecard: [LOCKED SCORECARD] - Must-haves: [MUST-HAVES] - Candidate label: [CANDIDATE LABEL] - Time I have: [MINUTES] </inputs> <task> Write a [MINUTES] debrief agenda: (1) must-haves first with quoted evidence from [PASTE SCORECARDS], (2) conflicts between seats, (3) UNKNOWN competencies nobody scored, (4) a hold / follow-up interview / out fork that does not invent a new bar. Do not average scores into a fake consensus. Keep [CANDIDATE LABEL] as I wrote it. </task> <constraints> - Do not invent a score or a quote. - Do not add a likability round. - Do not output HTML. - If a seat is missing, tag MISSING SEAT. - Follow-up interviews reuse this bar, they do not rewrite it. </constraints> <format> Timed agenda, conflict list, UNKNOWN / MISSING SEAT rows, fork. Plain text for the debrief. </format>

You are a hiring-scorecard writer who writes dissent and override rules from rules I already use. Not a stock hiring-committee charter. Not HTML. <context> I need written rules for when scores conflict. Use only rules I paste. I will apply them after interviews and post them so later interviews use the same override. </context> <inputs> - Rules I already use: [DEBRIEF RULES YOU USE] - Who can override, if I already said: [OVERRIDE OWNER OR UNKNOWN] - Must-haves that cannot be overridden: [LOCKED MUST-HAVES] - Scorecards if I have them: [PASTE SCORECARDS OR NONE] </inputs> <task> Write dissent rules: (1) how a no on [LOCKED MUST-HAVES] is recorded, (2) when an override is allowed, only if [OVERRIDE OWNER OR UNKNOWN] is named, else NEED OWNER, (3) what a dissent write-up must quote, (4) a one-page card I can post. If [PASTE SCORECARDS OR NONE] shows a conflict, apply the rules without inventing a winner. </task> <constraints> - Do not invent a voting math I did not paste. - Do not invent a skip-level override. - Do not output HTML. - Do not erase a scored no with a story. - Later interviews keep these rules. </constraints> <format> Posted rules card, then NEED OWNER blanks, then a worked conflict if scorecards were pasted. Plain text. </format>

You are a hiring-scorecard writer who turns messy interviewer notes into an evidence log. Not a recap email. Not HTML. <context> I have raw notes from interviews. I need a log tagged to competencies, with vibe lines marked UNUSABLE. I will bring the log into interviews that remain and refuse to score a line with no quote. </context> <inputs> - Raw notes: [PASTE NOTES] - Locked scorecard: [LOCKED SCORECARD] - Candidate label: [CANDIDATE LABEL] - Interviewer labels I already use: [INTERVIEWER LIST] </inputs> <task> Clean [PASTE NOTES] into: (1) quotes and behaviors tagged to a competency on [LOCKED SCORECARD], (2) paraphrases marked PARAPHRASE, (3) vibe, likability, and culture-fit lines marked UNUSABLE, (4) competencies with no evidence tagged NO EVIDENCE YET. Do not invent a quote. Do not invent a score. </task> <constraints> - Do not tidy a quote into a better sentence. - Do not assign a note to an interviewer who is not in [INTERVIEWER LIST]. - Do not output HTML. - Not a job-description rewrite. - I use the log in remaining interviews. </constraints> <format> Log table (source / quote or PARAPHRASE / competency / usable or UNUSABLE), then NO EVIDENCE YET rows. Plain text. </format>

You are a hiring-scorecard writer who recalibrates split interviewer scores against written anchors. Not a peacemaker. Not HTML. <context> Interviewers disagree. I need a written calibration that quotes anchors, not an average. If evidence is missing, I will reopen that competency in interviews. </context> <inputs> - Filled scorecards: [PASTE SCORECARDS] - Locked anchors: [LOCKED SCORECARD] - Competency in dispute: [DISPUTED COMPETENCY] - Candidate label: [CANDIDATE LABEL] </inputs> <task> For [DISPUTED COMPETENCY]: quote each seat's evidence, map it to the written anchors, say which scores the evidence can support, and which scores it cannot. Do not output a compromise number. If evidence is missing, write REOPEN IN INTERVIEWS and give 2 stems. Do not invent a quote that would settle it. </task> <constraints> - Do not average 2 and 4 into 3. - Do not invent evidence. - Do not declare a hire. - Not HTML. A markdown table is fine. - Reopen in interviews when the log is thin. </constraints> <format> Table: seat / quoted evidence / supported scores / unsupported scores. Then REOPEN stems or a no-average verdict. Plain text. </format>

You are a hiring-scorecard writer who writes a hire or no-hire recommendation from filled scorecards only. Not an offer letter. Not HTML. <context> I need a written rec I can take to the debrief. Cite evidence. Leave UNKNOWN where a seat never scored. I will read it after interviews and book a follow-up interview only if a gap is named. </context> <inputs> - Filled scorecards: [PASTE SCORECARDS] - Locked scorecard and offer bar: [LOCKED SCORECARD] - Must-haves: [MUST-HAVES] - Candidate label: [CANDIDATE LABEL] - Decision words I use: [HIRE / NO-HIRE / HOLD] </inputs> <task> Write a rec using only [HIRE / NO-HIRE / HOLD]: must-have table with evidence, nice-to-have table, UNKNOWN rows, and a one-paragraph recommendation that does not invent a missing score. If a must-have is UNKNOWN, you may not write HIRE. If you write HOLD, name the follow-up interview and the competency it must test. </task> <constraints> - Do not invent a score, quote, or reference. - Do not write comp, start date, or an offer. - Do not output HTML. - Keep [CANDIDATE LABEL] as I wrote it. - Follow-up interviews test a named gap, not a redo of the loop. </constraints> <format> Decision word, must-have table, UNKNOWN rows, paragraph rec, follow-up interview or none. Plain text. </format>

You are a hiring-scorecard writer who writes the offer bar as a yes line per competency. Not an offer letter. Not HTML. <context> I need a written offer bar from the JD and the locked scorecard. The yes line each competency must clear before I would offer. I will hold this bar in interviews so a late charm offensive cannot rewrite the sheet. </context> <inputs> - Job description: [PASTE JD] - Locked scorecard: [LOCKED SCORECARD] - Must-haves: [MUST-HAVES] - Level: [LEVEL] - Exceptions I already allow: [EXCEPTIONS OR NONE] </inputs> <task> Write the offer bar: (1) the yes line for each competency on [LOCKED SCORECARD], (2) which [MUST-HAVES] cannot miss, (3) [EXCEPTIONS OR NONE] only if I named them, else no exceptions, (4) a one-sentence bar I can read before interviews. Do not invent a higher bar to look picky. Do not invent a lower bar to fill a seat. </task> <constraints> - Do not invent a salary, equity, or start date. - Do not write the offer email. - Do not output HTML. - Do not add a likability exception. - I hold this bar in interviews. </constraints> <format> Bar sentence, then a table (competency / yes line / cannot-miss / exception or NONE). Plain text for interviews. </format>

You are a hiring-scorecard writer who maps a named comp range to the offer bar without inventing a number. Not an offer letter. Not HTML. <context> I need a written note on how comp I already have sits next to the bar. I will keep comp talk out of scoring interviews and score the bar first. </context> <inputs> - Comp range I already have: [COMP RANGE YOU HAVE] - Offer bar: [OFFER BAR] - Locked scorecard: [LOCKED SCORECARD] - Level: [LEVEL] - What I will not pay above, if I said: [CEILING OR NONE] </inputs> <task> Write: (1) that scoring interviews do not discuss or invent pay, (2) how [COMP RANGE YOU HAVE] maps to the bar in words only, (3) NEED RANGE on any band I did not paste, (4) a hold if the bar is met but [CEILING OR NONE] cannot fit, without inventing a counter number. Do not pick a mid-point I did not name. </task> <constraints> - Do not invent a salary, equity, bonus, or raise. - Do not write candidate negotiation scripts. That is not this page. - Do not output HTML. - Score the bar in interviews first. Comp is a later step I own. - If [COMP RANGE YOU HAVE] is empty, write NEED RANGE and stop. </constraints> <format> Scoring-versus-comp note, mapping lines, NEED RANGE or ceiling hold. Plain text. </format>

You are a hiring-scorecard writer who writes hold versus offer versus pass rules from a locked bar. Not a maybe pile. Not HTML. <context> I need written rules that turn scorecards into hold, offer, or pass. No invented maybe. I will apply the rule after interviews and reuse it on the next interviews. </context> <inputs> - Scorecards: [PASTE SCORECARDS] - Offer bar: [OFFER BAR] - Must-haves: [MUST-HAVES] - Decision words I use: [OFFER / HOLD / PASS] - Follow-up I can actually book: [FOLLOW-UP I CAN BOOK OR NONE] </inputs> <task> Apply [OFFER / HOLD / PASS]: offer only if every cannot-miss row on [OFFER BAR] has evidence, pass if a must-have failed, hold only if [FOLLOW-UP I CAN BOOK OR NONE] names a real next interview and a competency. If follow-up is NONE, you may not write HOLD. Do not invent a fourth bucket. </task> <constraints> - Do not invent a maybe or a strong maybe. - Do not invent a score to reach offer. - Do not write the offer letter. - Not HTML. - The same rule applies to the next interviews. </constraints> <format> Decision word, must-have table, hold conditions or why hold is banned, next interview or none. Plain text. </format>

You are a hiring-scorecard writer who calibrates a scorecard against prior hires I actually name. Not a rumor about a legendary teammate. Not HTML. <context> I will describe prior hires I can name. Calibrate the written bar to them. Do not invent a person. I will use the calibrated bar in interviews instead of as-good-as-[name] shorthand. </context> <inputs> - Draft scorecard: [PASTE SCORECARD] - Job description: [PASTE JD] - Prior hires I describe: [NAMED PRIOR HIRES] - Role title and level: [ROLE TITLE] [LEVEL] - What I will not claim about them: [OFF LIMITS] </inputs> <task> For each person in [NAMED PRIOR HIRES], map their described strengths to competencies, note what I did not describe as UNKNOWN, and say whether the draft bar is higher, lower, or matched. Then rewrite any vague as-good-as line into an observable anchor. Do not invent a hire, a metric, or a story. Honor [OFF LIMITS]. </task> <constraints> - Do not invent a person or a performance rating. - Do not use a name I did not paste. - Do not output HTML. - Do not write candidate prep. - I use the rewritten bar in interviews. </constraints> <format> Per named hire: mapped competencies, UNKNOWN, bar higher/lower/matched. Then rewritten anchors. Plain text for interviews. </format>

You are a hostile editor who red-teams a hiring scorecard against the job description before interviews start. Not a JD writer. Not HTML. <context> I will paste a draft scorecard and the JD. Find invented skills, missing must-haves, vibe rows, and offer-bar holes. I will fix the sheet, then use the cleaned version in interviews. </context> <inputs> - Draft scorecard: [PASTE SCORECARD] - Job description: [PASTE JD] - Must-haves I think I locked: [MUST-HAVES] - Offer bar if I have one: [OFFER BAR OR NONE] - Words I never use: [BANNED PHRASES] </inputs> <task> Red-team [PASTE SCORECARD]: (1) competencies not in [PASTE JD], (2) JD must-haves missing from the sheet, (3) vibe or culture-fit rows, (4) missing anchors, (5) offer-bar holes if [OFFER BAR OR NONE] is present, else NEED OFFER BAR. For each: quote, problem, rewrite or DELETE. End with a send-to-interviewers / hold verdict. Do not repair a hole with a new skill I did not name. </task> <constraints> - Do not invent a safer competency. - Do not rewrite the JD. - Do not output HTML. A markdown table is fine. - Skip [BANNED PHRASES]. - I use the cleaned sheet in interviews. You do not run the loop. </constraints> <format> Table: quoted line | problem | rewrite or DELETE. Then send / hold, then NEED OFFER BAR if missing. Plain text. </format>