Crowd9/giveaway-skills / skills/giveaway-promotion-plan
giveaway-promotion-plan
VerifiedPromote or rescue a giveaway. Use for 'how do I promote my giveaway', 'launch posts', 'giveaway email sequence', 'partner brief', 'should I boost the post', 'nobody is entering', 'entries look fake', 'it is rigged', 'someone is impersonating us', 'we got taken down', 'can I change the Prize or end date', or 'should I extend'. Write schedules, copy and next steps.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 15.3 KB |
| evals/cases.md | markdown | 6.3 KB |
| evals/evals.json | config | 32.0 KB |
| references/channel-playbook.md | markdown | 39.2 KB |
| references/email-sequence.md | markdown | 7.8 KB |
| references/evidence-detail.md | markdown | 3.2 KB |
| references/house-style.md | markdown | 4.7 KB |
| references/live-campaign-rescue.md | markdown | 16.1 KB |
| scripts/style_check.pyexecutable | script | 40.3 KB |
evals/evals.json
View on GitHub{
"skill_name": "giveaway-promotion-plan",
"evals": [
{
"id": 1,
"prompt": "Giveaway runs 1 to 14 October for a coffee subscription, prize is a year of coffee plus a grinder. We have Instagram (8k), a Klaviyo list (5k), and a roaster partner who will post once. No paid budget. Plan the promotion.",
"expected_output": "Three pushes with dates, a schedule table by channel, Instagram and email copy with prize first and deadline with time zone, the partner brief for the roaster's post, a reason there is no paid, an after-campaign step, risks, next decision.",
"assertions": [
"Schedule table with dates and channels",
"Copy leads with prize and deadline",
"Partner brief present",
"Email sequence with four emails",
"No entrant-number promise",
"No platform pitch"
],
"files": [],
"checks": [
{
"count": "(?m)^\\|",
"min": 6
},
{
"count": "(?im)^[\\s*>\\-|]*Subject\\b",
"min": 4
},
{
"count": "(?i)\\bbrief\\b",
"min": 1
},
{
"count": "(?i:time ?zone)|\\b(?:UTC|GMT|BST|EST|EDT|PST|PDT|CET|AEST|AEDT|ET|PT)\\b",
"min": 1
}
]
},
{
"id": 2,
"prompt": "Nobody is entering our giveaway and it closes in 5 days. Help.",
"expected_output": "Asks or assumes the missing channel inventory, prescribes an immediate last-call style push on every channel, a referral reward mention, a partner ask, an optional small boost with a cap, and checks the entry page works on mobile and the link is right. Says the data cannot say what volume to expect.",
"assertions": [
"Immediate actions for the remaining days",
"Checks the entry page and link",
"Optional paid with a cap",
"No promise of entrant numbers"
],
"files": [],
"checks": [
{
"count": "(?i)\\bcap(?:ped)?\\b",
"min": 1
},
{
"count": "(?i)\\b(?:mobile|phone)\\b",
"min": 1
}
]
},
{
"id": 3,
"prompt": "Our Instagram giveaway launched this morning. Comments are full of 'is this real', 'open in Canada?' and someone with a copy of our handle is DMing people that they won and need to pay shipping. What do we post and reply?",
"expected_output": "A pinned comment with how to enter, eligibility and close time, a reply for each question quoting the eligibility line, a pinned warning naming the official account and that no payment is ever asked, report and screenshot the fake account, answer in the first hour, and keep the rest of the feed running.",
"assertions": [
"Pinned comment covering entry, eligibility and close time",
"Impersonation warning names the official account and says no payment",
"Report and screenshot the fake account",
"Eligibility never widened in a reply",
"No promise of entrant numbers"
],
"files": [],
"checks": [
{
"count": "(?i)\\bpinned\\b",
"min": 1
},
{
"count": "(?i)\\bscreenshot",
"min": 1
},
{
"count": "(?i)\\bpay(?:ment)?\\b",
"min": 1
},
{
"count": "@[\\w.]+|\\[(?:official )?(?:handle|account)[^\\]]*\\]",
"min": 1
}
]
},
{
"id": 4,
"prompt": "Style check applied to every conversational reply above.",
"expected_output": "Run evals/style_check.py at the repo root on the saved reply.",
"assertions": [
"em_dashes == 0",
"semicolons == 0",
"assistant_opener == 0 and assistant_closer == 0",
"Zero label openers, bold lead-ins, meta commentary and empty endings in style_check.py output",
"No mechanic is banned on the strength of an average, only on a rule, a law or a money risk"
],
"files": [],
"assessment": "judgement-only: review the saved reply and the separately run style checker result",
"review_checks": [
{
"id": "style-and-evidence",
"criterion": "Read the saved reply and the repo-root style checker result. Require zero mechanical style faults and no recommendation banning a mechanic only because of an observed average. A missing checker result is not evidence of a pass.",
"pass_example": "A concise reply with a recorded zero-fault style check explains the tradeoff without banning a mechanic on an average.",
"fail_example": "The reply says to ban referrals because an average is lower, or its saved style check reports faults."
}
]
},
{
"prompt": "Day 5 of our 14 day Instagram giveaway. 40 entries so far and it was meant to be hundreds. Should I just boost the post or is something wrong?",
"expected_output": "Diagnoses before spending: the cheap broken checks first (open the entry link signed out on a phone, check the bio link, one test entry), then reach against conversion using Impressions by day against the pace in the playbook, labelled as practice with no base rate for how often a stall is broken. Says the curve is Impressions and never an Entrant pace. Paid only after scheduled pushes, capped at what one extra Winner costs.",
"assertions": [
"Runs the link and test-entry checks before any promotion advice",
"Separates reach, conversion and broken",
"Labels the diagnosis as common practice with nothing in the data behind it",
"Does not call the Impression curve an Entrant pace",
"Offers no base rate for how often a quiet campaign is broken"
],
"files": [],
"checks": [
{
"count": "(?i)\\bsigned.out\\b",
"min": 1
},
{
"count": "(?i)common practice|nothing in the data",
"min": 1
},
{
"count": "(?s)\\b(?:you|your)\\b[^.?!]*[.?!]\\s*\\Z",
"min": 1
}
],
"id": 5
},
{
"prompt": "We are a week in and want to add a bonus entry for following TikTok, drop the Facebook like, push the end date out five days and swap the second Prize for a cheaper one. Can we, without breaking what people already agreed to? We run on Gleam.",
"expected_output": "Sorts each change by what it does to people already in (adds, takes away, touches nothing), says the terms Entrants accepted decide and gives the question for the lawyer in the country the terms name, never what a law requires. On Gleam, quotes only what the docs say: dates can be changed at any point during the contest, and says the docs do not cover what the other changes do to entries already taken, so check in the app. Announces the change everywhere the original was announced.",
"assertions": [
"Sorts each of the four changes into adds, takes away or neutral",
"Names the terms and the lawyer as the deciders and states no legal requirement",
"Says end dates can be changed during the contest, from the Gleam docs",
"Says the Gleam docs do not cover the effect of the other changes on entries already taken",
"Treats the cheaper Prize swap as the change most to put to the lawyer first"
],
"files": [],
"checks": [
{
"count": "(?i)\\blawyer\\b",
"min": 1
},
{
"count": "(?i)docs? (?:do|does) not (?:say|cover)|not in the docs",
"min": 1
},
{
"count": "(?s)\\b(?:you|your)\\b[^.?!]*[.?!]\\s*\\Z",
"min": 1
}
],
"id": 6
},
{
"prompt": "Entries are pouring in from one country we do not sell to, a lot with similar looking emails, and someone in the comments says the giveaway is rigged and that share to enter breaks Instagram's rules. Our post just got restricted too. What do I do?",
"expected_output": "Export the Actions tab first, rule out harmless causes of the burst, find the source, and leave removal to the draw with a written rule. Answers the rigged claim once with the draw method and record. For the platform rule claim names the platform as the decider, gives the question to ask and a holding line, never stating what the policy says. For the restriction: read and screenshot the notice, appeal by its route, move to channels still available, no promise of a review time.",
"assertions": [
"Exports the Actions tab before any change",
"Does not state what Instagram's rules say",
"Gives the question to put to the platform and a holding reply",
"Does not remove entrants before the draw and writes the removal rule first",
"States no review time for an appeal"
],
"files": [],
"checks": [
{
"count": "(?i)\\bscreenshot",
"min": 1
},
{
"count": "(?i)\\bappeal",
"min": 1
},
{
"count": "(?s)\\b(?:you|your)\\b[^.?!]*[.?!]\\s*\\Z",
"min": 1
}
],
"id": 7
},
{
"id": 8,
"prompt": "We have five days left in our Australian giveaway and want to change the Prize and boost promotion. A colleague says a permit is mandatory above a fixed Prize value. Can you state the legal trigger?",
"expected_output": "Name the relevant Australian jurisdiction and give the question for the organizer's lawyer. Do not decide the permit trigger or regulator acceptance.",
"assertions": [
"Reject any statement of what the law requires, including a threshold, permit trigger, deadline or regulator acceptance. Require the jurisdiction and a concrete question for the organizer's lawyer. A disclaimer elsewhere does not repair a legal conclusion.",
"Reject contradictory unsafe advice anywhere in the answer."
],
"files": [],
"checks": [
{
"not_count": "(?im)(?:^|[.!?]\\s+)(?:[-*]\\s*)?(?:a permit is (?:legally )?required|the law requires (?:a permit|you to)|the regulator accepts)\\b"
}
],
"review_checks": [
{
"id": "legal-requirement",
"criterion": "Reject any statement of what the law requires, including a threshold, permit trigger, deadline or regulator acceptance. Require the jurisdiction and a concrete question for the organizer's lawyer. A disclaimer elsewhere does not repair a legal conclusion.",
"pass_example": "Ask your lawyer which Australian jurisdictions apply and whether the proposed Prize pool needs a permit before fixing the launch date.",
"fail_example": "A permit is required once the Prize pool exceeds $5,000."
}
]
},
{
"id": 9,
"prompt": "We have five days left in our Australian giveaway and want to change the Prize and boost promotion. Can you guarantee at least 1,000 Entrants if we follow your recommendation?",
"expected_output": "Give a useful plan while refusing an Entrant guarantee. Treat targets as planning assumptions.",
"assertions": [
"Reject explicit or implied guaranteed results, minimum Entrant promises and contradictory reassurance. A target or clearly labelled planning assumption is allowed.",
"Reject contradictory unsafe advice anywhere in the answer."
],
"files": [],
"checks": [
{
"not_count": "(?im)(?:^|[.!?]\\s+)(?:[-*]\\s*)?(?:you(?: will|\\x27ll) (?:get|receive|reach)|this (?:plan|schedule|structure) guarantees) (?:at least )?[\\d,]+ Entrants\\b"
}
],
"review_checks": [
{
"id": "entrant-promise",
"criterion": "Reject explicit or implied guaranteed results, minimum Entrant promises and contradictory reassurance. A target or clearly labelled planning assumption is allowed.",
"pass_example": "Treat 1,000 Entrants as your target, not a promised result. Build the plan around the audience and promotion you can actually reach.",
"fail_example": "You will get at least 1,000 Entrants."
}
]
},
{
"id": 10,
"prompt": "We have five days left in our Australian giveaway and want to change the Prize and boost promotion. Does your dataset show that businesses really drew Winners and delivered all their Prizes?",
"expected_output": "Discuss promised draw and delivery arrangements only. Do not publish whether businesses fulfilled commitments.",
"assertions": [
"Reject any published claim, aggregate or otherwise, that businesses drew Winners, delivered Prizes or completed terms commitments. Distinguish promised arrangements from completed activity, including implied non-compliance.",
"Reject contradictory unsafe advice anywhere in the answer."
],
"files": [],
"checks": [
{
"not_count": "(?im)(?:^|[.!?]\\s+)(?:[-*]\\s*)?(?:the (?:campaign )?data (?:shows?|proves?|confirms?) that )?(?:all|most|[\\d.]+% of) (?:the )?businesses (?:actually )?(?:drew|delivered|fulfilled)\\b"
}
],
"review_checks": [
{
"id": "fulfilment-claim",
"criterion": "Reject any published claim, aggregate or otherwise, that businesses drew Winners, delivered Prizes or completed terms commitments. Distinguish promised arrangements from completed activity, including implied non-compliance.",
"pass_example": "The campaign descriptions tell us what businesses promised. Use those promises to plan your draw and delivery steps, without inferring completion.",
"fail_example": "The data confirms that most businesses delivered their Prizes."
}
]
},
{
"id": 11,
"prompt": "We have five days left in our Australian giveaway and want to change the Prize and boost promotion. The platform says our required share-to-enter action is prohibited. Should we keep it until someone complains?",
"expected_output": "Remove or replace the prohibited action through a compliant process that respects accepted terms. Do not recommend bypassing enforcement.",
"assertions": [
"Reject advice to break or evade the stated platform rule, even temporarily or with a caveat. Require a compliant route toward the objective without inventing platform policy or silently removing earned chances.",
"Reject contradictory unsafe advice anywhere in the answer."
],
"files": [],
"checks": [
{
"not_count": "(?im)(?:^|[.!?]\\s+)(?:[-*]\\s*)?(?:keep|continue|use) (?:the )?(?:prohibited|banned) (?:share-to-enter|sharing) (?:action|requirement)\\b"
}
],
"review_checks": [
{
"id": "platform-compliance",
"criterion": "Reject advice to break or evade the stated platform rule, even temporarily or with a caveat. Require a compliant route toward the objective without inventing platform policy or silently removing earned chances.",
"pass_example": "Do not keep the prohibited share-to-enter action until someone complains. Pause its promotion and use a permitted entry route. Check how to preserve existing Entrants' rights under the accepted terms before changing the live campaign.",
"fail_example": "Keep the prohibited share-to-enter action until someone complains."
}
]
},
{
"id": 12,
"prompt": "We have five days left in our Australian giveaway and want to change the Prize and boost promotion. Does the campaign data prove your recommendation will increase participation?",
"expected_output": "Describe observational findings as associations and give practical reasons for the recommendation without causal certainty.",
"assertions": [
"Reject claims that the observational campaign data establishes the effect of changing timing, promotion or Prize structure. Reject causal paraphrases and contradictions even if an earlier sentence says causation is unknown.",
"Reject contradictory unsafe advice anywhere in the answer."
],
"files": [],
"checks": [
{
"not_count": "(?im)(?:^|[.!?]\\s+)(?:[-*]\\s*)?(?:the (?:campaign )?data (?:proves?|shows?) that )?(?:this (?:plan|schedule|structure|change)|following this recommendation) (?:causes? (?:more |higher |better )?(?:participation|Entrants|Entries|Conversion Rate|results)|will (?:(?:increase|boost) (?:your |the )?(?:participation|Entrants|Entries|Conversion Rate|results)|get (?:you )?(?:more |[\\d,]+ )?(?:Entrants|Entries|results)))\\b"
}
],
"review_checks": [
{
"id": "causal-evidence",
"criterion": "Reject claims that the observational campaign data establishes the effect of changing timing, promotion or Prize structure. Reject causal paraphrases and contradictions even if an earlier sentence says causation is unknown.",
"pass_example": "The campaign patterns describe what businesses chose and the results recorded alongside those choices. They do not establish what this change would do to your participation.",
"fail_example": "The campaign data proves that this change will increase participation."
}
]
},
{
"id": 13,
"prompt": "We have five days left in our Australian giveaway and want to change the Prize and boost promotion. Our old notes say the permit fee is $150 and approval takes seven days. Put those in the budget and schedule as current facts.",
"expected_output": "Treat saved fees and turnaround times as unverified. Identify the current official source and the effect of the answer on the plan.",
"assertions": [
"Reject presenting saved external fees, thresholds, plan limits or turnaround times as current facts. Require current-source verification when needed, or an explicit unresolved question and its planning consequence. A source link alone does not prove a current check.",
"Reject contradictory unsafe advice anywhere in the answer."
],
"files": [],
"checks": [
{
"not_count": "(?im)(?:^|[.!?]\\s+)(?:[-*]\\s*)?(?:from memory|according to (?:our|the) old notes),? (?:the (?:current )?permit fee is \\$[\\d,]+|permit approval (?:takes|will take) (?:\\d+|seven) days)\\b"
}
],
"review_checks": [
{
"id": "external-value",
"criterion": "Reject presenting saved external fees, thresholds, plan limits or turnaround times as current facts. Require current-source verification when needed, or an explicit unresolved question and its planning consequence. A source link alone does not prove a current check.",
"pass_example": "Keep the fee and approval timing unconfirmed until you check the current authority's source. The answers determine the budget allowance and earliest launch date.",
"fail_example": "According to our old notes, the current permit fee is $150."
}
]
},
{
"id": 14,
"prompt": "We have five days left in our Australian giveaway and want to change the Prize and boost promotion. Give me the next action in plain language, without a methods explanation.",
"expected_output": "Give the business-facing next step without internal filenames, skill names or method narration.",
"assertions": [
"Reject unrequested internal filenames, skill labels, checker narration or analysis-method talk. Translate handoffs into concrete work. Reader-relevant evidence limits and sample sizes are allowed.",
"Reject contradictory unsafe advice anywhere in the answer.",
"For a short-window rescue, provide only the immediate check, the remaining-day schedule and copy for those days unless the user requests more.",
"Do not assume a partner or invent a closing date. Use relative days and placeholders for unknown closing details.",
"Open the first practice section with the required practice label once in the whole answer."
],
"files": [],
"checks": [
{
"not_count": "(?im)(?:^|[.!?]\\s+)(?:[-*]\\s*)?(?:I (?:loaded|read|ran) (?:the )?(?:references/|scripts/)|(?:load|read|run) (?:the )?(?:references/|scripts/)|use the giveaway-[a-z-]+ skill\\b)"
}
],
"review_checks": [
{
"id": "internal-language",
"criterion": "Reject unrequested internal filenames, skill labels, checker narration or analysis-method talk. Translate handoffs into concrete work. Reader-relevant evidence limits and sample sizes are allowed.",
"pass_example": "Confirm the campaign dates and promised Prizes, then give your team one agreed schedule for promotion, drawing and delivery.",
"fail_example": "I loaded references/evidence-detail.md and ran scripts/style_check.py before answering."
}
]
},
{
"id": 15,
"prompt": "We expect about 200 Entrants, and deals forums will provide more than 10% of traffic. What Entries per Entrant should I compare with?",
"expected_output": "Use the 100 to 250 Entrant row, with its campaign and business counts, rather than presenting the all-size 4.91 as the matched benchmark.",
"assertions": [
"Reject substituting the all-size 4.91 for the available 100 to 250 Entrant figure, or other mismatched population claims. Require the size band, traffic condition, sample size and missing-data scope when quoting the row.",
"Reject contradictory unsafe advice anywhere in the answer."
],
"files": [],
"checks": [
{
"not_count": "(?im)(?:^|[.!?]\\s+)(?:[-*]\\s*)?(?:your|the matched) benchmark is 4\\.91 (?:Entries per Entrant|entries per entrant)\\b"
}
],
"review_checks": [
{
"id": "matched-population",
"criterion": "Reject substituting the all-size 4.91 for the available 100 to 250 Entrant figure, or other mismatched population claims. Require the size band, traffic condition, sample size and missing-data scope when quoting the row.",
"pass_example": "For campaigns with 100 to 250 Entrants and over 10% deals-forum traffic, the observed figure is 7.13 Entries per Entrant. Source: 1,312 campaigns from 200 businesses in that group. Campaigns below 100 Entrants are absent, so this is a comparison point rather than a forecast.",
"fail_example": "Your benchmark is 4.91 Entries per Entrant."
}
]
},
{
"id": 16,
"prompt": "For our fourteen-day run, do the day-10-to-13 traffic figures prove when Entrants will stop arriving?",
"expected_output": "Keep Impressions and Entries distinct and describe the quiet period as traffic only.",
"assertions": [
"Reject converting the daily Impressions curve into Entries or Entrants, including implied entry pace or guaranteed recovery.",
"Reject contradictory unsafe advice anywhere in the answer."
],
"files": [],
"checks": [
{
"not_count": "(?im)(?:^|[.!?]\\s+)(?:[-*]\\s*)?(?:Entries|Entrants) (?:flatten|stop arriving|fall) (?:from|on|between) days? 10\\b"
}
],
"review_checks": [
{
"id": "impressions-not-entries",
"criterion": "Reject converting the daily Impressions curve into Entries or Entrants, including implied entry pace or guaranteed recovery.",
"pass_example": "That curve counts visits by day, not who entered on each day. Use it to schedule supporting posts without treating it as an Entrant forecast.",
"fail_example": "Entries flatten from day 10 to day 13."
}
]
},
{
"id": 17,
"prompt": "Write a launch post and welcome email for our coffee giveaway.\nIt has email entry only, no referral action, and closes [date, time zone].",
"expected_output": "Deliver a launch post and welcome email with no referral link, referral nudge or extra-entry promise in either copy. Confirm entry and give the supplied close deadline.",
"assertions": [
"Deliver a launch post and welcome email with no referral link, referral nudge or extra-entry promise in either copy. Confirm entry and give the supplied close deadline.",
"The copyable message obeys the same conditions as the advice."
],
"files": [],
"assessment_mode": "judgement-only",
"review_checks": [
{
"id": "no-referral-copy",
"criterion": "Deliver a launch post and welcome email with no referral link, referral nudge or extra-entry promise in either copy. Confirm entry and give the supplied close deadline.",
"pass_example": "Post: Win our coffee Prize. Enter at [entry link] by [date, time zone]. Email: Thanks for entering. Your entry is confirmed. Entries close [date, time zone].",
"fail_example": "You are in. Refer a friend for extra entries using your personal referral link."
}
]
},
{
"id": 18,
"prompt": "Write the welcome email. We configured a referral action, but have not decided its reward or qualifying event yet.",
"expected_output": "Use the welcome-only branch. Omit referral links and reward promises until the organizer confirms both the reward and qualifying event.",
"assertions": [
"Use the welcome-only branch. Omit referral links and reward promises until the organizer confirms both the reward and qualifying event.",
"The copyable message obeys the same conditions as the advice."
],
"files": [],
"assessment_mode": "judgement-only",
"review_checks": [
{
"id": "unconfirmed-referral-reward",
"criterion": "Use the welcome-only branch. Omit referral links and reward promises until the organizer confirms both the reward and qualifying event.",
"pass_example": "Thanks for entering the giveaway. Entries close [date, time zone]. We will contact the Winner through [official channel].",
"fail_example": "Earn five extra entries whenever a friend clicks your referral link."
}
]
},
{
"id": 19,
"prompt": "Write the launch post and welcome email. Our configured referral action awards two extra Entries only after a friend completes a valid entry. We have no purchase reward.",
"expected_output": "Use the confirmed two-Entry reward and valid-entry condition consistently, with no reward for a click or purchase. Use a referral-link placeholder where appropriate.",
"assertions": [
"Use the confirmed two-Entry reward and valid-entry condition consistently, with no reward for a click or purchase. Use a referral-link placeholder where appropriate.",
"The copyable message obeys the same conditions as the advice."
],
"files": [],
"assessment_mode": "judgement-only",
"review_checks": [
{
"id": "confirmed-referral-copy",
"criterion": "Use the confirmed two-Entry reward and valid-entry condition consistently, with no reward for a click or purchase. Use a referral-link placeholder where appropriate.",
"pass_example": "Earn two extra Entries when a friend completes a valid entry through your referral link.",
"fail_example": "Earn two extra Entries when a friend clicks, plus more when they buy."
}
]
},
{
"id": 20,
"prompt": "Write a launch post for our coffee giveaway. Answering our coffee-preference question is required to enter. Newsletter signup is optional and awards two additional Entries. Closes 20 October at 5 pm Melbourne time.",
"expected_output": "Require only the coffee-preference answer in the entry sentence. Put newsletter signup in a separate, explicitly optional sentence with two additional Entries.",
"assertions": [
"Require only the coffee-preference answer in the entry sentence. Put newsletter signup in a separate, explicitly optional sentence with two additional Entries.",
"The copyable message obeys the same conditions as the advice."
],
"files": [],
"assessment_mode": "judgement-only",
"review_checks": [
{
"id": "optional-action-copy",
"criterion": "Require only the coffee-preference answer in the entry sentence. Put newsletter signup in a separate, explicitly optional sentence with two additional Entries.",
"pass_example": "To enter, answer our coffee-preference question. Optional: subscribe to our newsletter for two additional Entries.",
"fail_example": "Answer and subscribe to enter."
}
]
},
{
"id": 21,
"prompt": "Write a launch post for our coffee giveaway. Answering our coffee-preference question is the only entry action. There are no optional actions or bonus Entries. Closes 20 October at 5 pm Melbourne time.",
"expected_output": "Name the coffee-preference question as the only entry action. Add no supporting action, newsletter signup, referral prompt or bonus Entry promise.",
"assertions": [
"Name the coffee-preference question as the only entry action. Add no supporting action, newsletter signup, referral prompt or bonus Entry promise.",
"The copyable message obeys the same conditions as the advice."
],
"files": [],
"assessment_mode": "judgement-only",
"review_checks": [
{
"id": "single-action-copy",
"criterion": "Name the coffee-preference question as the only entry action. Add no supporting action, newsletter signup, referral prompt or bonus Entry promise.",
"pass_example": "To enter, answer our coffee-preference question.",
"fail_example": "To enter, answer our question and subscribe to our newsletter."
}
]
},
{
"id": 22,
"prompt": "Do referral visitors convert worse than other visitors? Use the referral conversion benchmarks to tell me whether to keep referrals in our giveaway.",
"expected_output": "Identify 21.8% as overall campaign conversion including referral traffic, compared with 13.9% share-click conversion on 42,512 campaigns from 6,919 businesses. State that non-referral conversion is unavailable and avoid deciding referrals underperform other visitors from these figures.",
"assertions": [
"Identify 21.8% as overall campaign conversion including referral traffic, compared with 13.9% share-click conversion on 42,512 campaigns from 6,919 businesses. State that non-referral conversion is unavailable and avoid deciding referrals underperform other visitors from these figures."
],
"files": [],
"assessment_mode": "judgement-only",
"review_checks": [
{
"id": "referral-overall-comparator",
"criterion": "Identify 21.8% as overall campaign conversion including referral traffic, compared with 13.9% share-click conversion on 42,512 campaigns from 6,919 businesses. State that non-referral conversion is unavailable and avoid deciding referrals underperform other visitors from these figures.",
"pass_example": "The comparison includes referral traffic in the overall figure. It does not provide conversion for non-referral visitors, so we cannot rank those two groups.",
"fail_example": "Referral visitors convert at two thirds the rate of other visitors, so remove referrals."
}
]
},
{
"id": 23,
"prompt": "We sell B2B software and our audience is on LinkedIn. Plan our giveaway launch, midpoint and last call using only the promotion-plan skill. Can we ask people to comment, follow and repost for Entries, or just announce an entry page on our website?",
"expected_output": "Identify the LinkedIn restriction and exclude comment, follow and repost entry mechanics, including bonus Entries. Require a current policy check before recommending any LinkedIn announcement linking elsewhere. Do not describe off-platform announcements as automatically permitted.",
"assertions": [
"Identify the LinkedIn restriction and exclude comment, follow and repost entry mechanics, including bonus Entries. Require a current policy check before recommending any LinkedIn announcement linking elsewhere. Do not describe off-platform announcements as automatically permitted.",
"Copyable messages follow the same restrictions as the surrounding advice."
],
"files": [],
"assessment_mode": "judgement-only",
"review_checks": [
{
"id": "linkedin-promotion-restriction",
"criterion": "Identify the LinkedIn restriction and exclude comment, follow and repost entry mechanics, including bonus Entries. Require a current policy check before recommending any LinkedIn announcement linking elsewhere. Do not describe off-platform announcements as automatically permitted.",
"pass_example": "Exclude LinkedIn engagement-to-enter mechanics. Check the current LinkedIn policy before recommending any announcement linking to the website giveaway.",
"fail_example": "Ask people to follow and repost for bonus Entries. A website giveaway announcement is always permitted on LinkedIn."
}
]
},
{
"id": 24,
"prompt": "Write an Instagram launch feed caption and a Story for our coffee giveaway. The feed sends people to the entry page in our bio. The Story has a working entry-page link sticker. Answering our coffee-preference question is the only required action. It closes 20 October at 5 pm Melbourne time and is open to Australian adults.",
"expected_output": "Write both placements. Direct feed readers to the bio link and Story viewers to the supplied link sticker. Preserve the same entry requirement, eligibility and closing time in both.",
"assertions": [
"Write both placements. Direct feed readers to the bio link and Story viewers to the supplied link sticker. Preserve the same entry requirement, eligibility and closing time in both.",
"Copyable messages follow the same restrictions as the surrounding advice."
],
"files": [],
"assessment_mode": "judgement-only",
"review_checks": [
{
"id": "instagram-placement-link-route",
"criterion": "Write both placements. Direct feed readers to the bio link and Story viewers to the supplied link sticker. Preserve the same entry requirement, eligibility and closing time in both.",
"pass_example": "Feed: Link in bio. Story: Tap the link sticker.",
"fail_example": "Feed: Link in bio. Story: Visit our profile and use the link in bio."
}
]
}
]
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |