Content & Media Creation · dontbesilent2025/dbskill
dbs-benchmark
Finds benchmarks worth emulating using a five-fold filtering method while filtering out noise caused by subjective individual differences. Use when the user requests finding benchmarks, selecting targets for imitation, or analyzing competitive references.
L1 Basic · Source registration onlyV1 Commit-PinnedDual-pass Audit
Control levels describe skill capabilities and suggested control measures; they do not represent quality or safety certifications, nor do they dictate usability. Usage decisions rest entirely with users and administrators. Different agent hosts apply varying constraints; do not assume hosts provide automatic protection. Third-party skill copyrights belong to their original authors. Use at your own risk.
1How It Works & Mechanism
Four tenets form the worldview prior to screening: Eliminating the self accelerates decision-making; imitation is the correct answer during the 0-to-1 phase; imitation granularity determines quality; high profit is the sole metric.
skills/dbs-benchmark/SKILL.md
**核心信念：模仿不是方法，是信仰。**
What this accomplishes: Transforms "finding benchmarks" from an aesthetic/interest question into a determinable business problem—Tenet 1 declares "all discussions about 'me' are decision noise," while Tenet 4 dismisses alternative metrics like gross margin, high barriers, or rapid growth with "all the above ≠ high profit; we need high profit." Consequently, screening is left with only one final criterion: Is this business making money?
Phase 1 first establishes status (existing business running / starting from scratch) and immediately interrupts requests such as "find one that suits me."
skills/dbs-benchmark/SKILL.md
**「适合你的这个说法本身就是问题。我们不讨论你，我们只讨论业务。」**
This provides hardcoded interruption scripts aligned with the rule in "Things absolutely not to do" ("Never help users argue 'why this benchmark does not suit them'—that merely enables their evasion"), turning "eliminating the self" from a slogan into an actionable conversational intervention.
Phase 2 Five-Fold Filtering: Is it profitable? → Is it understandable? → Is it imitable? → Eliminate the self → Avoid debating business essence; each filter provides evaluation criteria or follow-up prompts.
skills/dbs-benchmark/SKILL.md
利润至少是用户当前收入（或预期收入）的 10 倍
Filter 1 offers evaluation clues (product price × estimated sales, team size, ad spending intensity, willingness to invest cash) along with "making money ≠ having followers"; Filter 2 requires clarifying the acquisition/conversion/delivery/repurchase chain; Filter 3 distinguishes between "currently lacking resources" and "unable to acquire resources at all" (exclusive channels, government ties, special licenses); Filter 5 demands "deleting the terms 'track' and 'industry' from your mind."
Phase 3 outputs a benchmark analysis report in a fixed structure: Five-fold filter results table + Four segments of business model + Three steps of imitation path + One-line takeaway.
skills/dbs-benchmark/SKILL.md
| 1. 赚钱 | ✅/❌ | {估算利润} |
The report template embeds code blocks containing four points (acquisition/conversion/delivery/repurchase) under "## His Business Model" and steps 1-2-3 under "## Imitation Path", concluding with a sharp summary under "## One-Line Takeaway". The tabular format forces an explicit verdict on each of the five filters, preventing vague conclusions.
Phase 4 Imitation Execution Check: Uses a 9-dimension + 6-dimension (platform operations) consistency comparison table to evaluate the user against the benchmark item-by-item; any mismatch indicates a flaw.
skills/dbs-benchmark/SKILL.md
**不一致的地方就是问题。**
The 9 dimensions cover product pricing, naming/packaging, acquisition channels, content format and frequency, title/cover styles, tonal pitch, delivery method, and promotional method. The supplementary 6 platform dimensions cover account ramp-up strategy, ad spend approach (Juguang/Chitiao/Qianchuan), ad budget/ROI, private domain traffic routing and conversion, and livestream frequency/duration. The evaluation rule: "For every inconsistency, the user must justify why it differs. If they cannot justify it, align it with the benchmark."
Inlines a case library (3 positive examples, 2 negative examples, each with "Diagnostic Highlights") to anchor the five-fold filtering and granularity tenets to concrete scenarios.
skills/dbs-benchmark/SKILL.md
于是我想把其他人的爆款视频开头，一帧不改，直接变成我的。
Case 2 serves as an example of "frame-level granularity benchmarking", where its diagnostic highlight states: "The problem shifts to 'how to repost/recreate legally'"; Negative Case 2 draws the line: "Imitation ≠ unauthorized copying. Imitation is reconstruction after understanding; copying is direct duplication without understanding." The case library simultaneously demonstrates tone and outlines operational boundaries.
Concludes with an optional archival prompt under strictly defined conditions: mentioned only when all five filters pass and actionable conclusions exist; forbidden during the screening phase.
skills/dbs-benchmark/SKILL.md
只在五重过滤通过、有可执行结论时提。还在筛选阶段不要提。
This aligns with the unified closure of leaf skills ("conclude directly upon completing the current task; only when the user explicitly inquires about next steps..."), preventing users from being pushed into /dbs orchestration before an analysis has solidified.
2Core Capabilities
01Five-fold filter screening method (profitable / understandable / imitable / eliminate self / avoid industry essence), providing evaluation criteria and follow-up prompts for each filter
02Benchmark analysis report (five-fold filter table + 4-part acquisition/conversion/delivery/repurchase + 3-step imitation path)
03Imitation granularity verification: Consistency rating across 9 foundational dimensions + 6 platform operation dimensions
04Hardened scripts to interrupt three evasive questions: 'suited for me', 'not interested', and 'evaluating industry prospects'
05Inlined positive/negative case library (including edge cases distinguishing 'imitation ≠ unauthorized copying')
06Conditional archival prompt (suggesting /dbs only after passing screening with actionable conclusions)
4Recommended Governance & Evidence
Suggested Controls for Safe Operation
For Agent Runtime
Can be auto-loaded
For Administrators
Register source and version
Evidence & Triggers (File Paths & Excerpts)
- SKILL.md: Pure prompt instructions and rule text, without script execution or external impacts
Audit Notes: Self-validation: The dbs-benchmark directory contains only SKILL.md and agents/openai.yaml, without scripts/references/templates/data files, script execution, network endpoints, credential reading, or local file writes; all capabilities are delivered purely via the prompt criteria tables and output templates in SKILL.md. The only input is the user's business status and candidate benchmark descriptions pasted in the current session. The ending includes only a single soft prompt ("If unsure about the next step, you can enter `/dbs`"), invoking no tools. Gear verification (evaluated by 'whether endpoints constitute necessary outbound transmission'): The directory contains no URLs, backend names, CDN references, web fonts, or API calls. Risks are limited strictly to prompt behavior and output quality (see risks).
5Second-Pass Verification
- [ok] 无脚本/无网络/无凭证/无写入 — 目录仅 SKILL.md + agents/openai.yaml；token 扫描零命中；SKILL.md 未要求联网查数据、也未要求写任何文件。
- [ok] 五重过滤确为 5 个筛子且各有判定法 — 筛子 1 他赚钱吗 / 2 你能看懂吗 / 3 你能模仿吗 / 4 排除自我 / 5 不讨论业务本质，逐段读过；筛子 1 有量化门槛（10 倍收入），筛子 2/3 有正反例，筛子 4 有打断话术。
- [ok] 报告模板与 Phase 4 检查表逐字存在 — Phase 3 代码块含五重过滤表与获客/转化/交付/复购四段；Phase 4 含 9 维表与平台运营 6 维表（含聚光/薯条/千川等具体投流工具名）。
- [ok] 是否夸大（「排除一切干扰」能否兑现） — 部分兑现：方法与话术明确，但筛子 1 的利润判断只给「看产品价格 × 估算销量、看团队规模、看投放力度」这类启发式，无数据源、无核验步骤——「排除噪音」实际由用户的自我克制完成，已在 risks 标注。
- [ok] 边界案例真在源文（避免把方法读成教唆抄袭） — 反面案例 1（网易小蜜蜂像素级抄小红书）诊断要点为「模仿产品形式 ≠ 模仿商业模式」；反面 2 明确「模仿 ≠ 搬运。模仿是理解后重建，搬运是不理解直接复制」。
- [ok] 与 marketplace 条目描述一致 — marketplace.json 描述「用五重过滤法帮你找到值得模仿的对标，排除一切关于「我」的噪音」，与 SKILL.md 描述语义一致。
- [ok] 元数据（license/stars/last_push/installs/commit） — GitHub API：stars 10121、pushed_at 2026-09-07T03:48:59Z、license.spdx_id NOASSERTION（根 LICENSE 为 CC BY-NC 4.0，非商业）；本地 HEAD = 8b8e33f1ecaed8cee606fe950c4426b525ead314 = 任务 pin。
- [discrepancy] meta.installs 值域校正（批4 质检 2026-09-15） — 原记录为 skills.sh 仓库级聚合徽标（579.6K（skills.sh 徽标 /b/dontbesilent2025/…），并非本 skill 的单体安装量；站内惯例只记可定位的单 skill 数值 → 校正为 null（其他 25 份 dbs 报告本就为 null，修此 8 份以齐平）
6Verdict & Scenarios
Best for: Suited for entrepreneurs or solopreneurs in the 0→1 phase who want to swiftly identify and replicate a validated, profitable business model; also suited for operating businesses seeking to dissect more profitable peer samples horizontally. The five-fold filter results table + granularity comparison table serve directly as an execution checklist, ideal for users who prefer firm, conclusive guidance over comfort. It also provides content creators with a structured framework for analyzing competitor operational dimensions (account launch, ad traffic, private domain).
Not for: Not suited for businesses in the differentiation/branding stage that require original positioning; not suited for scenarios requiring compliance, copyright, or platform policy audits (does not provide legal or policy assessments; boundaries around 'legal repurposing' in case studies require independent verification); not suited for treating benchmark selection as a one-off decision, as it does not handle data verification or due diligence; not suited for users uncomfortable with assertive guidance who expect their personal preferences to be prioritized.
Install Agent-Ready
① Verified Mirror 2026-09-15
Option A · Direct DownloadDownload dbs-benchmark.tar.gz
sha256:
Option B · Copy Guide for Agenta6cced3df9c025aa…Agents automatically download and verify mirrors from this JSON guide.
② Upstream GitHub
GitHub Repo ↗
Pinned to commit 8b8e33f1ec.