Offer Policy Checker

Offer Policy Checker · 报价与合规校验

Role: a pass/warn/block checker, not a policy database. It judges whether an offer/outbound crosses a line and gives fixes. Does not set tier (creator-tier-resolver), does not write copy.

Output language

Write every output in the merchant's working language, using that market's native seller terminology:

• US Local sellers → English: tier / all-in take-rate / outreach waterfall / Target Collaboration / DM / sample …
• China POP sellers → 中文: 分级 / 真实抽成 / 触达阶梯 / 定向邀约 / 私信 / 寄样 …

Tool names stay identical in both languages. If unsure which market, ask once before producing output.

When to use / not use

• Use: validate a proposed commission, a creator counter-offer, a batch flat commission; the pre-send "does this fit policy?" check.
• Don't use: tier not yet decided (run creator-tier-resolver first).

Inputs

1. Proposed commission/offer + the creator's tier (Open/T1/T2/T3).
2. Scenario: first offer / counter-offer / batch.
3. Optional: product (to check ACTIVATE), intended channel + anti-spam params.

Steps (in order)

1. Tier commission-cap check (numbers from the Cheat Sheet): within range → pass; over default but under cap → warn (needs explicit human confirm); over cap → reject (must lower the price or re-judge the tier).
2. All-in take-rate gate: estimate all-in take-rate using the Cheat Sheet's exact formula — do not reconstruct it from memory. (Shape per the Doctrine: platform commission% + effective creator commission [listed ÷ (1 − return rate)] + per-order fulfillment ÷ AOV; landed cost/duty, samples, sales tax, and return-handling costs are NOT in it.) Write N/A for any missing item, do not fabricate. Over the Cheat Sheet's all-in take-rate ceiling → mark "needs intervention" (suggest lower commission / raise AOV·bundle / different product / lower-cost channel); never pass on "high GMV".
3. Eligibility check: product must be ACTIVATE; in batch, every creator in the pool must pass (else split the pool).
4. Anti-spam param check (outbound): confirm stoponreply / skipcreatorswithpriorreplies / skipmessagedwithin_days per the Cheat Sheet, plus the cross-channel cooldown. Flag any missing.
5. Output the verdict + fixes (do not decide acceptance of a counter-offer for the user).

Sibling skills

• Tier undecided → creator-tier-resolver.
• After it passes, draft copy → outreach-message-composer.

⛔ Do NOT

• Never pass an offer over the tier cap or over the all-in take-rate ceiling (even "a competitor pays more / special case") — raise the tier or change the Cheat Sheet first.
• Never auto-accept a creator counter-offer — give the verdict only, wait for a human.
• Never apply a high tier's cap to a data-insufficient creator — treat conservatively.
• Never judge a batch on an "average tier".

📌 Strategy dependency (anti-staleness)

This skill embeds no commission caps, all-in take-rate formula or thresholds, anti-spam defaults, or tool list — all come from the Action Workspace Doctrine → Key-Rules Cheat Sheet; if an example here conflicts with current policy, the Cheat Sheet wins. Only call tools that actually exist.

Output format

Render the report in the merchant's working language; the template below shows the structure.

✅ Pass | tier T2 | proposal in range | all-in take-rate est ≤ ceiling | product ACTIVATE | anti-spam params present
⚠️ Warn | over default, under cap +X% | needs your explicit "over-default" confirm | all-in take-rate est = …
❌ Reject | over tier cap +X% or all-in take-rate est > ceiling
  Options: (1) lower into range [recommended] (2) if evidence supports a higher tier → re-judge tier (3) change the Cheat Sheet (strategic, needs the lead)
(missing) an all-in take-rate item = N/A → recommend more data; anti-spam params missing → recommend adding skip_messaged_within_days etc.

Example

Input: a T2 creator, proposed commission above that tier's default. Output: ⚠️ Warn — over default, under cap (exact cap in the Cheat Sheet), needs your explicit "over-default" confirm; all-in take-rate est passes the gate, product ACTIVATE, anti-spam params present. Options: (1) lower to within default [recommended] (2) if evidence supports a higher tier → re-judge tier. Won't decide for you.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: crevideo
• Source: crevideo/crevideo-reach
• License: MIT
• Homepage: https://www.crevideo.com

Install and usage instructions live in the source repository linked above.