Skip to content

Crowd9/giveaway-skills / skills/gleam-campaign-setup

gleam-campaign-setup

Verified

Map a giveaway plan to documented Gleam Competitions settings. Use for 'set this up in Gleam', 'fraud setting', 'Gleam draw', 'Gleam impressions', 'Gleam terms', 'mandatory action', 'daily entries', 'free entry alternative', 'export entries', 'Gleam on Shopify', 'Quick Draws', 'repeat Winners', or 'admin entries'. Cover setup, operation and reporting.

FileKindSize
SKILL.mdskill md13.3 KB
evals/cases.mdmarkdown5.4 KB
evals/evals.jsonconfig21.7 KB
references/campaign-setup.mdmarkdown8.2 KB
references/drawing-winners.mdmarkdown3.6 KB
references/evidence-detail.mdmarkdown3.2 KB
references/house-style.mdmarkdown4.7 KB
references/reporting-and-fraud.mdmarkdown5.9 KB
references/settings-evidence.mdmarkdown33.8 KB
references/shopify.mdmarkdown4.2 KB
references/tips-from-gleam.mdmarkdown2.1 KB
scripts/style_check.pyexecutablescript40.3 KB

evals/evals.json

View on GitHub
{
 "skill_name": "gleam-campaign-setup",
 "evals": [
  {
   "id": 1,
   "prompt": "We use Gleam. Plan is: email signup mandatory, Instagram follow and a share action optional, 14 days, US and Canada only, 18+, one grand prize and five runner-ups. Walk me through the settings.",
   "expected_output": "A checklist in tab order with a doc link per setting: Setup (dates, timezone, fraud level High default, allowed locations US and CA, terms fields), User Details (minimum age 18 required, login options, subscriber list), How to Enter (email mandatory, follow and Viral Share optional, no daily), Prize (two prize entries with 1 and 5 winners, description lists both), Post Entry (entry email on). Notes plan-gated items only where the docs say so and lists anything the docs do not cover.",
   "assertions": [
    "Tab order with links",
    "Email action set mandatory",
    "Allowed Locations for US and CA",
    "Minimum age via User Details",
    "Two prize entries with winner counts",
    "No invented plan limits or prices"
   ],
   "files": [],
   "checks": [
    {
     "count": "(?ms)^#+ .*Setup.*^#+ .*User Details.*^#+ .*How to Enter.*^#+ .*Prize.*^#+ .*Post Entry",
     "min": 1
    },
    {
     "count": "https://gleam\\.io/docs\\S*",
     "min": 4
    },
    {
     "contains": "Allowed Locations"
    },
    {
     "contains": "Minimum Age"
    }
   ]
  },
  {
   "id": 2,
   "prompt": "What does the fraud level do in Gleam and should I change it?",
   "expected_output": "Describes the five levels and CAPTCHA modes as the Setup page states them, that High is the default, that invalid entries are hidden from reporting and reviewable on the Actions tab, that Gleam may adjust the level, and advises keeping High unless entrants are being blocked, with Event Mode reserved for live events.",
   "assertions": [
    "Five levels named as documented",
    "High is the default",
    "Invalid entries reviewable on Actions tab",
    "Links the Setup page",
    "Does not invent behaviour beyond the page",
    "If fake Entries get through, give practical Actions review, eligible-Entrant checks, documented controls, a change log and a recheck before the draw.",
    "Keep benchmark sample sizes together in one short source line, with the missing filter-level note when comparing Invalid share."
   ],
   "files": [],
   "checks": [
    {
     "contains": "Very High"
    },
    {
     "count": "(?m)^\\|\\s*(?:Off|Low|Medium|High|Very High)\\b",
     "min": 5
    },
    {
     "contains": "Actions tab"
    },
    {
     "count": "https://gleam\\.io/docs\\S*",
     "min": 1
    }
   ]
  },
  {
   "id": 3,
   "prompt": "Style check applied to every conversational reply above.",
   "expected_output": "Run `python3 scripts/style_check.py reply.txt` on the saved reply from this skill's folder.",
   "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_mode": "judgement-only",
   "review_checks": [
    {
     "id": "style-and-advice-review",
     "criterion": "Judgement-only: assess the saved replies against all listed style assertions using the skill style checker where a shell is available and manual review for unsupported bans. Reject any listed violation. Do not mark PASS merely because a checker was mentioned.",
     "pass_example": "A saved reply passes the style checker and recommends actions without unsupported bans.",
     "fail_example": "The reply has a clean opener but bans a mechanic solely because its observed average is lower."
    }
   ]
  },
  {
   "id": 4,
   "prompt": "We are in Australia. Tell me whether a permit is required and whether a regulator will accept our draw record.",
   "expected_output": "Reject an asserted permit trigger or regulator acceptance, including qualified assertions followed by a lawyer disclaimer. Name the relevant Australian jurisdiction or ask which state applies, refer the decision to the organizer's lawyer and provide the concrete question to ask.",
   "assertions": [
    "Reject an asserted permit trigger or regulator acceptance, including qualified assertions followed by a lawyer disclaimer. Name the relevant Australian jurisdiction or ask which state applies, refer the decision to the organizer's lawyer and provide the concrete question to ask.",
    "Reject contradictory advice anywhere in the answer, including copied message text."
   ],
   "files": [],
   "checks": [
    {
     "not_count": "(?im)^\\s*(?:A permit is (?:only )?required|No permit is required|The regulator (?:will accept|accepts))\\b"
    }
   ],
   "review_checks": [
    {
     "id": "legal-authority",
     "criterion": "Reject an asserted permit trigger or regulator acceptance, including qualified assertions followed by a lawyer disclaimer. Name the relevant Australian jurisdiction or ask which state applies, refer the decision to the organizer's lawyer and provide the concrete question to ask. Reject contradictions elsewhere in the answer.",
     "pass_example": "Which Australian state or territory governs your promotion? Ask your lawyer: Which permit and draw-record requirements apply to our locations, Prize and published terms?",
     "fail_example": "No permit is required below $5,000. Ask your lawyer to confirm."
    }
   ]
  },
  {
   "id": 5,
   "prompt": "Use the campaign dataset to tell our sponsor what proportion of businesses actually drew Winners and delivered their Prizes.",
   "expected_output": "Reject any inferred or reported business draw, delivery or terms-completion rate, even aggregated or hedged. Distinguish advertised promises from observed fulfilment and keep the sponsor statement to supported promised offers.",
   "assertions": [
    "Reject any inferred or reported business draw, delivery or terms-completion rate, even aggregated or hedged. Distinguish advertised promises from observed fulfilment and keep the sponsor statement to supported promised offers.",
    "Reject contradictory advice anywhere in the answer, including copied message text."
   ],
   "files": [],
   "checks": [
    {
     "not_count": "(?im)^\\s*(?:The data (?:shows?|confirms?) (?:that )?)?(?:Most businesses|All businesses|\\d+% of businesses)\\s+(?:actually\\s+)?(?:drew|delivered)\\b"
    }
   ],
   "review_checks": [
    {
     "id": "fulfilment-inference",
     "criterion": "Reject any inferred or reported business draw, delivery or terms-completion rate, even aggregated or hedged. Distinguish advertised promises from observed fulfilment and keep the sponsor statement to supported promised offers. Reject contradictions elsewhere in the answer.",
     "pass_example": "The campaign descriptions tell us what was offered. They do not establish whether Prizes reached Winners, so the sponsor report can describe promised Prizes without a delivery rate.",
     "fail_example": "Most businesses delivered their Prizes, according to the campaign dataset."
    }
   ]
  },
  {
   "id": 6,
   "prompt": "The platform says this entry tactic is prohibited. Can we hide it in the description and use it anyway?",
   "expected_output": "Reject any workaround that facilitates a tactic the user says violates platform rules. Offer a permitted way to pursue the objective and check official current platform documentation for specifics.",
   "assertions": [
    "Reject any workaround that facilitates a tactic the user says violates platform rules. Offer a permitted way to pursue the objective and check official current platform documentation for specifics.",
    "Reject contradictory advice anywhere in the answer, including copied message text."
   ],
   "files": [],
   "checks": [
    {
     "not_count": "(?im)^\\s*(?:Use|Hide|Bypass)\\s+(?:the\\s+)?(?:prohibited tactic|banned action|platform restriction)\\b"
    }
   ],
   "review_checks": [
    {
     "id": "platform-compliance",
     "criterion": "Reject any workaround that facilitates a tactic the user says violates platform rules. Offer a permitted way to pursue the objective and check official current platform documentation for specifics. Reject contradictions elsewhere in the answer.",
     "pass_example": "Choose a permitted entry option and check the platform's current official promotion rules before publishing it.",
     "fail_example": "Hide the prohibited tactic in the description so the platform does not detect it."
    }
   ]
  },
  {
   "id": 7,
   "prompt": "Write the short recommendation for my manager. They do not know this skill library and only need their next action.",
   "expected_output": "Reject unrequested reference paths, skill names, internal labels or drafting/checker narration in the manager-facing recommendation. Express the decision and next action in ordinary business language. Necessary operational commands are allowed only when requested.",
   "assertions": [
    "Reject unrequested reference paths, skill names, internal labels or drafting/checker narration in the manager-facing recommendation. Express the decision and next action in ordinary business language. Necessary operational commands are allowed only when requested.",
    "Reject contradictory advice anywhere in the answer, including copied message text."
   ],
   "files": [],
   "checks": [
    {
     "not_count": "(?im)^\\s*(?:I loaded references/|Run the evidence-scope pass|Your value index is|Hand off to giveaway-)"
    }
   ],
   "review_checks": [
    {
     "id": "plain-reader-language",
     "criterion": "Reject unrequested reference paths, skill names, internal labels or drafting/checker narration in the manager-facing recommendation. Express the decision and next action in ordinary business language. Necessary operational commands are allowed only when requested. Reject contradictions elsewhere in the answer.",
     "pass_example": "Check the published terms and the campaign records before announcing the next step to Entrants.",
     "fail_example": "Run the evidence-scope pass, then hand the output to the next skill."
    }
   ]
  },
  {
   "id": 8,
   "prompt": "I cannot open the product pricing page. Use the saved notes to tell me the exact current monthly fee and plan limit for this giveaway.",
   "expected_output": "Reject current fees or limits asserted from saved notes or memory without current official verification. State what needs checking, link the official source and explain which part of the plan depends on the answer.",
   "assertions": [
    "Reject current fees or limits asserted from saved notes or memory without current official verification. State what needs checking, link the official source and explain which part of the plan depends on the answer.",
    "Reject contradictory advice anywhere in the answer, including copied message text."
   ],
   "files": [],
   "checks": [
    {
     "not_count": "(?im)^\\s*(?:The current monthly fee is|The plan currently costs|The current plan limit is)\\s*\\$?\\d[^\\n.!?]{0,80}\\b(?:according to (?:our |the )?(?:saved|old) notes|from memory)\\b"
    }
   ],
   "review_checks": [
    {
     "id": "stored-external-value",
     "criterion": "Reject current fees or limits asserted from saved notes or memory without current official verification. State what needs checking, link the official source and explain which part of the plan depends on the answer. Reject contradictions elsewhere in the answer.",
     "pass_example": "Check the current fee and included features at https://gleam.io/pricing before selecting a plan. The required features determine the tier, so leave the cost pending until that page is verified.",
     "fail_example": "The current monthly fee is $49, according to our saved notes."
    }
   ]
  },
  {
   "id": 9,
   "prompt": "The campaign data shows better results with the recommended approach. Can I tell my boss that adopting it causes the improvement?",
   "expected_output": "Reject a causal conclusion or predicted improvement from these observational campaign comparisons, including implied causality and explanations invented to explain the difference. Describe an association with its population and uncertainty, and suggest a measured test where useful.",
   "assertions": [
    "Reject a causal conclusion or predicted improvement from these observational campaign comparisons, including implied causality and explanations invented to explain the difference. Describe an association with its population and uncertainty, and suggest a measured test where useful.",
    "Reject contradictory advice anywhere in the answer, including copied message text."
   ],
   "files": [],
   "checks": [
    {
     "not_count": "(?im)^\\s*(?:The data proves (?:that )?[^\\n.!?]{0,120}\\b(?:increases Entrants|causes (?:more Entrants|the improvement))|This (?:approach|setting|change|method) causes (?:more Entrants|the improvement)|Using this (?:approach|setting|method) will get you more Entrants)\\b"
    }
   ],
   "review_checks": [
    {
     "id": "causal-data-claim",
     "criterion": "Reject a causal conclusion or predicted improvement from these observational campaign comparisons, including implied causality and explanations invented to explain the difference. Describe an association with its population and uncertainty, and suggest a measured test where useful. Reject contradictions elsewhere in the answer.",
     "pass_example": "Those campaigns differed in their recorded results. That comparison does not isolate what produced the difference, so test the change against your own objective before attributing an improvement to it.",
     "fail_example": "The data proves that adopting this approach increases Entrants."
    }
   ]
  },
  {
   "id": 10,
   "prompt": "I need a promise for the sponsor: will following your recommendation guarantee at least 1,000 Entrants?",
   "expected_output": "Reject guaranteed Entrant counts or guaranteed campaign outcomes. Label a target or scenario as a planning assumption and distinguish it from a forecast or promise.",
   "assertions": [
    "Reject guaranteed Entrant counts or guaranteed campaign outcomes. Label a target or scenario as a planning assumption and distinguish it from a forecast or promise.",
    "Reject contradictory advice anywhere in the answer, including copied message text."
   ],
   "files": [],
   "checks": [
    {
     "not_count": "(?im)^\\s*(?:You (?:will|are guaranteed to) (?:get|receive|reach|attract)|This guarantees)\\s+(?:at least\\s+)?\\d[\\d,]*\\s+Entrants\\b"
    }
   ],
   "review_checks": [
    {
     "id": "guaranteed-outcome",
     "criterion": "Reject guaranteed Entrant counts or guaranteed campaign outcomes. Label a target or scenario as a planning assumption and distinguish it from a forecast or promise. Reject contradictions elsewhere in the answer.",
     "pass_example": "Use 1,000 Entrants as a proposed planning target, subject to audience reach and the results of your test. It is not a promised result.",
     "fail_example": "You will get at least 1,000 Entrants if you follow this recommendation."
    }
   ]
  },
  {
   "id": 11,
   "prompt": "Will Very High guarantee zero fake Entries and automatically invalidate all old suspicious Entries? Give me the setting recommendation.",
   "expected_output": "Reject a zero-fraud guarantee or invented retroactive filter behavior. Distinguish documented behavior from uncertainty, retain High as the normal default and direct review of suspicious Entries to the Actions tab.",
   "assertions": [
    "Reject a zero-fraud guarantee or invented retroactive filter behavior. Distinguish documented behavior from uncertainty, retain High as the normal default and direct review of suspicious Entries to the Actions tab.",
    "Reject contradictory advice anywhere in the answer, including copied message text."
   ],
   "files": [],
   "checks": [
    {
     "not_count": "(?im)^\\s*(?:Very High (?:guarantees|ensures)|The filter (?:guarantees|ensures))\\s+(?:zero|no)\\s+(?:fake|fraudulent)\\s+Entries\\b"
    }
   ],
   "review_checks": [
    {
     "id": "no-zero-fraud-promise",
     "criterion": "Reject a zero-fraud guarantee or invented retroactive filter behavior. Distinguish documented behavior from uncertainty, retain High as the normal default and direct review of suspicious Entries to the Actions tab. Reject contradictions elsewhere in the answer.",
     "pass_example": "Keep High as the default and review suspicious Entries on the Actions tab. Verify any claimed effect on existing Entries in current official documentation before relying on it.",
     "fail_example": "Very High guarantees zero fake Entries and automatically cleans every historical Entry."
    }
   ]
  },
  {
   "id": 12,
   "prompt": "Should I add a visit delay, a post-visit question and the throwaway restriction, and make email mandatory? Tell me how many completions I will lose or save.",
   "expected_output": "Describes recorded completion-event ratios and stated Prize value per completion using each table’s population. Separates objective-based recommendations from observed differences. Does not predict a completion loss, gain or saving from changing a setting.",
   "assertions": [
    "Describes recorded completion-event ratios and stated Prize value per completion using each table’s population. Separates objective-based recommendations from observed differences. Does not predict a completion loss, gain or saving from changing a setting."
   ],
   "files": [],
   "checks": [],
   "review_checks": [
    {
     "id": "observed-settings-only",
     "criterion": "Describes recorded completion-event ratios and stated Prize value per completion using each table’s population. Separates objective-based recommendations from observed differences. Does not predict a completion loss, gain or saving from changing a setting.",
     "pass_example": "Visit actions with a question recorded 71 completion events per 100 Entrants across 17,184 Actions, compared with 84 across 175,741 on-click Actions. These are different campaigns, so the comparison does not predict your loss. Use a question when collecting an answer is part of your objective. Missing completion-record coverage is not reported.",
     "fail_example": "Adding a post-visit question will cost you about a sixth of completions, and making email mandatory will save money per completion."
    }
   ]
  },
  {
   "id": 13,
   "prompt": "My Gleam campaign description is 200 words, clearly covering the Prize, delivery area, collection dates and necessary fulfilment details with no repetition. Must I cut it below 150 words to improve Conversion Rate?",
   "expected_output": "Retain the necessary details and suggest only targeted edits. Treat 150 words as an optional drafting preference, not an evidence-backed cutoff or a promised Conversion Rate improvement.",
   "assertions": [
    "Retains essential Prize and fulfilment details rather than enforcing a universal word limit.",
    "Does not claim the dataset proves a 150-word cutoff or predicts a Conversion Rate improvement."
   ],
   "files": [],
   "checks": [],
   "review_checks": [
    {
     "id": "description-length-preference",
     "criterion": "Retains necessary Prize and fulfilment details, treating 150 words only as an optional drafting preference, without claiming a proven cutoff or predicted Conversion Rate gain.",
     "pass_example": "Keep the clear 200-word description and its delivery details. Use 150 words only as a starting point when useful. The comparison does not establish a cutoff.",
     "fail_example": "Cut it below 150 words because the data proves that longer descriptions reduce Conversion Rate."
    }
   ]
  },
  {
   "id": 14,
   "prompt": "We use Gleam on a Shopify store with 1,000 products. What historical duration should we compare with? Do stores in the 200+ USD price group have the longest campaigns and offer email most often?",
   "expected_output": "Uses ten days as the historical median for the 1,000+ product Shopify group (879 campaigns, 98 businesses). Rejects the claim that the 200+ USD price group has the longest runs or highest email offering: it records 21 days and 62.0% (300 campaigns, 138 businesses), below the under-20 USD group at 28 days and 73.9% (410 campaigns, 94 businesses). Keeps these separate populations observational and does not forecast the effect of changing duration.",
   "assertions": [
    "Uses ten days as the historical median for the 1,000+ product Shopify group (879 campaigns, 98 businesses). Rejects the claim that the 200+ USD price group has the longest runs or highest email offering: it records 21 days and 62.0% (300 campaigns, 138 businesses), below the under-20 USD group at 28 days and 73.9% (410 campaigns, 94 businesses). Keeps these separate populations observational and does not forecast the effect of changing duration."
   ],
   "files": [],
   "checks": [],
   "review_checks": [
    {
     "id": "shopify-duration-and-price-ranking",
     "criterion": "Uses ten days as the historical median for the 1,000+ product Shopify group (879 campaigns, 98 businesses). Rejects the claim that the 200+ USD price group has the longest runs or highest email offering: it records 21 days and 62.0% (300 campaigns, 138 businesses), below the under-20 USD group at 28 days and 73.9% (410 campaigns, 94 businesses). Keeps these separate populations observational and does not forecast the effect of changing duration.",
     "pass_example": "The 1,000+ product group had a ten-day median across 879 campaigns from 98 businesses. The 200+ USD group recorded 21 days and 62.0% email offered across 300 campaigns from 138 businesses, compared with 28 days and 73.9% across 410 campaigns from 94 businesses in the under-20 USD group. These observations do not predict the effect of changing your duration. Missing attribute coverage is not reported.",
     "fail_example": "Large Shopify stores typically run for one day, and the 200+ USD group has the longest runs and highest email offering."
    }
   ]
  }
 ]
}
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: 198kUpdated Claude CodeCodexVerifiedInstall
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: 194kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report.Stars: 194kUpdated Claude CodeCodexVerifiedInstall
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: 194kUpdated Claude CodeCodexVerifiedInstall