本文へ移動
cccskills
無料GitHub で公開

case-camouflage-skill

Use when Amazon Ads optimization analysis, recommendations, approval packets, demos, recordings, or public artifacts must preserve exact source KPIs while masking account/profile/project names, products, ASINs, SKUs, campaigns, ad groups, keywords, search terms, targets, placements, filenames, URLs, and source-derived identifiers. Produces masked approval packets only; it must not directly mutate Amazon Ads.

インストール方法を見る

含まれるファイル(11)

  • SKILL.md4.2 KB
  • agents/openai.yaml431 B
  • references/approvals.md798 B
  • references/artifact-log-safety.md791 B
  • references/codename-resolution.md543 B
  • references/coverage.md840 B
  • references/lifecycle.md566 B
  • references/masking.md1.2 KB
  • references/metrics.md984 B
  • references/registry.md1.1 KB
  • references/style.md597 B

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Case Camouflage Skill

Use this skill when preparing Amazon Ads optimization output that may be shown to users, reviewers, public repos, demos, recordings, docs, or evals. The job is to camouflage the case context: keep optimization math real and exact while masking only user-facing labels and source-derived identifiers.

Non-Negotiables

  • Preserve exact KPIs and optimization logic. Do not fake, redact, perturb, swap, incorrectly merge, or round metrics for privacy.
  • Analyze first with raw source IDs. Group, rank, attribute, diagnose, and decide before masking. Never group by masked handles.
  • Mask display-plane labels and identifiers: account/profile/project names, products, ASINs, SKUs, campaigns, ad groups, keywords, search terms, targets, placements, filenames, URLs, and source-derived identifiers.
  • Use stable tenant/profile-scoped handles such as ACCOUNT-000001, PROFILE-000001, PRODUCT-000001, ASIN-000001, CAMPAIGN-000001, KW-000001, and TARGET-000001.
  • Do not expose registry mappings, real customer data, source IDs, credentials, raw reports, private execution manifests, HMAC digests, or raw API readbacks in public output.
  • Do not directly mutate Amazon Ads. You may create masked approval packets and private execution manifests for a separate approved execution tool after explicit approval.

Workflow

  1. Confirm scope: tenant/profile, marketplace, date windows, available reports, requested output surface, and whether this is public/demo/recording output.
  2. Load or configure a tenant-scoped masking registry. If text-only identifiers are present, require a per-tenant HMAC secret. See registry.
  3. Run the optimization in the analytical plane using raw source IDs and exact metrics. See metrics.
  4. Resolve display handles only after analysis. See masking.
  5. Build a masked approval packet for recommended actions. Keep private execution manifests in ignored private paths only. See approvals.
  6. Scan public artifacts, logs, stdout/stderr, rationales, readbacks, metadata, hidden sheets, and filenames before release. See artifact/log safety.
  7. Report coverage with counts and statuses only, never mappings. See coverage.

Reference Map

  • Masking: what must be masked, handle format, and display-plane rules.
  • Metrics: exact KPI preservation and source-plane grouping.
  • Registry: providers, collisions, unsafe aliases, placeholders, planned reservations, activation, HMAC.
  • Approvals: approval packets, action IDs, private manifests, stale packets.
  • Lifecycle: analysis to approval to readback status flow.
  • Artifact And Log Safety: scanners and public artifact release gates.
  • Codename Resolution: replacing customer/project codenames with generic language.
  • Coverage: safe summaries and limitations.
  • Style: output tone and table requirements.

Required Output Shape

For user-facing results, include:

  • Scope and mode: masked_output, read-only, no Amazon Ads mutation.
  • Data coverage and limitations.
  • Source-plane diagnostic findings with masked handles and exact KPIs.
  • Approval packet rows with action_id, masked profile/entity handles, current/proposed values, exact metrics, risk, preflight, readback, and monitoring.
  • Execution gate: separate approved execution tool required; this skill did not write to Amazon Ads.
  • Leak-scan result and registry coverage summary.

If a raw label or identifier appears in a draft, stop and remediate before showing it.

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Act as a senior Amazon account growth operator that combines Amazon performance-drop analytics, Amazon growth-opportunity analytics, Amazon Ads data, retail readiness, BSR/rank signals, margin, inventory, and campaign/search-term evidence into a prioritized profitable growth action plan. Use when Codex needs to decide what to protect, fix, scale, reduce, pause, harvest, launch, or monitor for an Amazon brand, agency, marketplace, ASIN set, or account; when building daily, weekly, or monthly Amazon account operating plans; when reallocating ad budget; when deciding which ASINs are safe or risky to push; or when turning performance-drop and growth-opportunity findings into approval-gated actions with success/failure criteria.

日本語の概要は準備中です。原文の説明を表示しています。

nospicyplease/amazon-ppc-advanced-skills142026年5月28日 更新

Diagnose why Amazon advertising performance declined for a product, ASIN, campaign, marketplace, or account. Use when the user asks why sales, orders, ROAS, ACOS, TACoS, profit, BSR, rank, traffic, conversion rate, or campaign performance dropped, asks for dropped campaigns, says investigate drop/drop analysis, or asks about wasted spend, negative keywords, negative targets, keywords, targets, or search terms in a drop context.

日本語の概要は準備中です。原文の説明を表示しています。

nospicyplease/amazon-ppc-advanced-skills142026年5月28日 更新

Replace this with a precise trigger description. State the Amazon PPC workflow, user intent, required data, and when Codex or Claude should use this skill.

日本語の概要は準備中です。原文の説明を表示しています。

nospicyplease/amazon-ppc-advanced-skills142026年5月28日 更新

Analyze Amazon product, sales, Amazon Ads, retail-readiness, and Best Seller Rank data to find commercially actionable growth opportunities. Use when Codex needs to identify which ASINs, campaigns, keywords, search terms, product targets, placements, categories, or BSR movements deserve more investment, optimization, harvesting, bid/budget changes, retail-readiness fixes, investigation, protection, or controlled scaling across Sponsored Products, Sponsored Brands, Sponsored Display, search term reports, targeting reports, campaign reports, placement reports, advertised/purchased product reports, Business Reports, organic performance, and BSR history.

日本語の概要は準備中です。原文の説明を表示しています。

nospicyplease/amazon-ppc-advanced-skills142026年5月28日 更新

Find Amazon Ads search terms ready for exact-match harvesting from auto, broad, phrase, or other discovery campaigns. Use when Codex needs standalone or Rocketcart MCP-aware search-term harvesting: classify terms by harvest readiness, resolve live IDs, choose safe destination campaigns/ad groups, preflight duplicates/current negatives/destination feasibility, decide whether source negatives are justified, execute only explicitly approved rows, and read back/monitor Sponsored Products or Sponsored Brands actions without unsafe negatives, duplicate routing, stale approvals, brand-defense cuts, own-ASIN-defense cuts, launch/rank mistakes, or low-sample overfitting.

日本語の概要は準備中です。原文の説明を表示しています。

nospicyplease/amazon-ppc-advanced-skills142026年5月28日 更新

Use for read-first Amazon Sponsored Products optimization, product-aware growth, preflight readiness, and post-change monitoring reviews that can run standalone from static exports or use Rocketcart MCP as an optional Amazon Ads plus product-intelligence layer. Guides Codex to inspect Rocketcart profiles, live SP campaigns, product ads/ASIN context, category rank/BSR movement, competitor signals, product readiness, budget and targeting drift, snapshots/changelogs, and proposed approval-gated action rows without executing writes by default.

日本語の概要は準備中です。原文の説明を表示しています。

nospicyplease/amazon-ppc-advanced-skills142026年5月28日 更新

nospicyplease のスキルをすべて見る

このスキルの問題を報告する