Hubbestaiskill.com

Home / All Skills / Content & Media Creation / dbs-content-system

Content & Media Creation · dontbesilent2025/dbskill

dbs-content-system

Organizes local drafts, posts, topics, case studies, and course materials into a sustainably growing, recombinable, and reusable content engineering system. Use when the user requests structuring content assets, organizing legacy materials, or establishing topic maps.

L4 High-Privilege · Graded auth & full auditV1 Commit-PinnedExecute CommandWrite FileDual-pass Audit

Author dontbesilent2025GitHub dontbesilent2025/dbskill ↗Stars 10121License CC BY-NC 4.0（Attribution-NonCommercial 4.0 International；仓库根 LICENSE 明文 NonCommercial — 非商业授权，不得商用；GitHub API license=NOASSERTION/Other，因 CC BY-NC 不在 SPDX 常见许可标识内；Copyright (c) 2026 dontbesilent）Commit 8b8e33f1ec

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

Single-directory heavy skill: The deliverable is not a summary, but a functioning local content engineering system ready for continuous operation; SKILL.md requires self-containment.

skills/dbs-content-system/SKILL.md

**你交付的不是一份总结，而是一套能继续运转的系统。**

The self-containment requirement is strictly defined: "**This skill must be self-contained. Do not assume the user can access knowledge packs, reference docs, or extra support files in the repository after installation. It must execute completely using only this single `SKILL.md`.**"

The directory self-maintains all assets: Scaffolds, templates, scripts, and documentation stay inside the skill directory and are distributed with installation (10 Node scripts + 7 templates + 4 root-level scaffolds + 6 rules + 2 documents).

skills/dbs-content-system/SKILL.md

**本 skill 不是轻量 prompt，而是单目录重型 skill。`SKILL.md`、脚手架、模板、脚本、文档都固定留在 `skills/dbs-content-system/` 目录内部，不依赖共享目录。**

SKILL.md lists these distributable files item by item and states: "If these files are missing from the installed skill package, delivery is considered incomplete."

Audit before building the project: Two threshold gates (volume and boundary); if not met, only audit findings and a downgrade path are provided.

skills/dbs-content-system/SKILL.md

### 原则 1：先审计，再建工程

Hardcoded numerical thresholds: No fewer than 50 processable text files, or no less than 80,000 extractable body characters; source scope must match at least 2 categories.

Raw materials are copied, never modified: Source files in the source directory remain untouched; copies are placed in `01-原始素材区/完整副本/` strictly as a trace-back reference.

skills/dbs-content-system/SKILL.md

原目录里的原文件不碰。

All processing occurs within the new project to prevent corrupting existing user materials.

Processing targets content units, not files: Initially fixed into 5 types (QST/CON/OPI/CAS/SOL), each as an independent Markdown file with YAML frontmatter.

skills/dbs-content-system/SKILL.md

首期只保留 5 类内容单元：

File rules require: "- File names must be fixed as `ID_Title.md`", and "- The current file represents the current valid version, leaving historical changes to Git".

Freeze criteria for relations and deduplication beforehand: Only 4 relation types (responds/explains/proves/conflicts) and 4 deduplication types are allowed, with only exact and synonymous duplicates merged by default.

skills/dbs-content-system/SKILL.md

第一期只允许 4 类关系：

The merge scope is bounded by "Only `Exact Duplicates` and `Synonymous Duplicates` are merged by default" to prevent accidental merging of near-duplicates.

Four-stage phased upgrade (Audit → Sample → Batch → Full), with hard gates per stage where missing a single condition blocks progression.

skills/dbs-content-system/SKILL.md

只有前一档闸门全部通过，才允许进入下一档。少一条都不升档。

Always "enters via `Audit Mode` by default"; promotion gates are listed point-by-point under headings like `Audit Mode → Sample Mode`.

Delivery tooling layer included with the skill: The init script builds the directory skeleton, copies templates and scaffolds, and copies 9 tool scripts into the project's `07-脚本与工具/`.

skills/dbs-content-system/SKILL.md

- `tools/init-content-system.js`：初始化工程骨架

Script evidence: Three loops using `const src = path.join(skillRoot, "templates", name);` / `fs.copyFileSync(src, dst);` and `const src = path.join(skillRoot, "tools", name);` copy templates/scaffolds/tools to the target project.

Initial sample extraction relies on a "fixed protocol + script assistance", not black-box automated semantic extraction.

skills/dbs-content-system/SKILL.md

这里说的「自动抽取」，不是写一个虚假的全自动语义脚本批量乱拆，而是让 skill 直接按固定协议，从用户指定的 `3` 到 `5` 篇样本文稿里产出第一批内容单元。

The protocol requires extracting 1 mandatory QST + 1 mandatory OPI per sample article, adding CON/CAS/SOL as needed, and states: "Do not pretend you can extract all semantic objects from a draft in one pass."

Structure precedes scale: First verify field/relation/deduplication stability using samples before batching; transitioning from batch to full has 7 additional gates (including misjudgments ≤3 in a 30-unit sample audit).

skills/dbs-content-system/SKILL.md

#### `批量模式 → 全量模式` 升档闸门

Sample-to-batch transition requires 0 missing relation validation targets and rebuildable state-layer files; batch-to-full requires 2 consecutive batches with no rule changes, plus "- Manual audit of `30` content units with no more than `3` major misjudgments".

"Functional" has an acceptable definition: Delivery can occur once the skeleton, rules, material copies, initial units, topic maps, assembly drafts, relation/deduplication indices, and state overview are complete, without committing to full completion.

skills/dbs-content-system/SKILL.md

- `03-处理状态/处理状态总览.md` 已明确当前范围、未处理量与下一步入口

Accompanied by a default quantity target of "- Produce at least `15` content units" and the minimum initial delivery standards in docs/acceptance.md.

Execution discipline requires actual execution: Do not stop at advice; point out insufficient materials immediately, narrow boundaries if ambiguous, and clearly define entry points to other skills.

skills/dbs-content-system/SKILL.md

- 不要停留在建议层

Entry points specify: "- `/dbs-good-question` has clearly formulated the problem specification and is suitable for automated execution"; it also specifies that this skill does not decide the next Skill on its own.

2Core Capabilities

01Audit threshold evaluation (Volume: 50 files/80,000 characters + 2 source types + inclusion/exclusion boundaries)

02Project skeleton initialization and asset distribution (init-content-system.js copies templates/scaffolds/tools)

03Raw material copies and three status registries (Source Registry / Raw Material Index / Backlog)

04Initial sample extraction protocol (3-5 articles, mandatory QST+OPI per article, enriched source/topic/keyword/relation metadata)

05Content unit specification (independent files, ID_Title.md, frontmatter, 7 minimal fields, 4 relation types, 4 deduplication types)

06Topic map and topic assembly draft generation (≥3 maps, ≥2 assembly drafts)

07Validation pipeline for relations, deduplication, and state indices (8 runnable commands)

08Four-stage phased upgrade gates and functional state verification

0910 Node utility tools distributed with the skill (init/source-registry/ledger/unit-draft/extract/assemble/link-map/duplicate/obsidian-links/summarize)

3External Dependencies

Kind	Dependency / Endpoint
cli	node
cli	npx skills（skills.sh CLI，安装途径）

4Recommended Governance & Evidence

Suggested Controls for Safe Operation

For Agent Runtime

Graded/per-use authorization, sandbox recommended

For Administrators

Approval workflow, rollback support, full audit

Evidence & Triggers (File Paths & Excerpts)

scripts/: Execute local scripts or commands: node tools/init-content-system.js <target_project_dir> (builds directory skeleton, writes status files, and copies templates/scaffolds)
本地工作目录: File I/O and artifact generation: Inside target project: `00-规则与索引/` `01-原始素材区/` `02-内容单元库/` `03-处理状态/` `04-模板/` `05-主题地图/` `0

Audit Notes: . The evaluation target is the skill's own directory: It includes 10 self-contained Node scripts (only require("fs")/require("path")) performing I/O operations on the local project such as creating directories, copying, writing files, and moving files to `.trash`; the entire directory has zero network endpoints, zero credential reads, zero child processes, and zero arbitrary code execution (ZERO HITS on dangerous tokens), thus falling into the blue tier (local scripts and local read/write only, no outbound network activity). Notable points: ① Scripts copy the user's source directory to the new project and write numerous files; the current working directory (cwd) and target project are determined by the user/agent, so running in the wrong directory may pollute the workspace; ② Deduplication cleanup uses rename into `.trash` instead of deletion, keeping files on disk; ③ extract-sample-units.js is a corpus-specific heuristic script rather than a general-purpose semantic extractor, requiring manual verification of false extractions.

5Second-Pass Verification

[ok] 自包含声明与资产面一致 — SKILL.md 列出的 templates/（7）、scaffold/root/（4）、scaffold/rules/（6）、docs/（2）、tools/（10）与 find 清点结果逐项吻合，无缺失文件。
[ok] 脚本仅本地读写、无网络无凭证 — 10 个脚本全部 `const fs = require("fs");` + `const path = require("path");`；grep https?://|fetch(|child_process|execSync|spawn|process.env|API_KEY 零命中。
[ok] 写盘范围限定在目标工程 — init 脚本所有目标路径均由 skillRoot/targetRoot 派生；extract 脚本的清理用 `fs.renameSync(filePath, target)` 移入 `<root>/.trash/{日期}_{原因}/`（不删除）。
[discrepancy] 抽取器非通用语义引擎（能力边界） — SKILL.md 声明「不是写一个虚假的全自动语义脚本批量乱拆」，但 extract-sample-units.js 内含大量作者语料特化正则（如 `if (/生产型兴趣/.test(context.conceptText) && /兴趣/.test(context.normalizedTitle) && /赚钱|变现/.test(context.normalizedTitle)) {`），换语料会退化到通用兜底分支；这不是「虚假」，但确属「对自有语料特化」，已在 risks 写明。
[ok] 四档闸门与数量门槛 — 审计/样本/批量/全量四模式与各自闸门标题均在原文；50 文件/80000 字、≥15 单元、≥3 地图/≥2 装配稿、抽查 30 误判≤3 逐字命中。
[ok] 源文件只读与副本制 — 「原目录里的原文件不碰。」+ 原则 3 标题「### 原则 3：原始素材不改写，只复制副本」逐字命中。
[ok] description 与内容相符 — description（把本地文稿/推文/选题/案例/课程稿整理成可持续生长、可重组复用的内容工程）对应五类单元 + 主题地图 + 装配稿 + 四模式推进，无夸大；但「可重组复用」的可验证证据是 assemble-topic-from-units.js 的 --auto 跨主题推荐，属启发式实现。
[ok] commit / license / 元数据 — HEAD = 8b8e33f1ecaed8cee606fe950c4426b525ead314；LICENSE = CC BY-NC 4.0（含 NonCommercial，非商业授权）；stars 10121、pushed_at 2026-09-07T03:48:59Z。

6Verdict & Scenarios

Truly deliverable tool-oriented skill: Not a methodology document, but a self-contained package of SKILL.md + templates + scaffolds + 10 scripts.

Gated progression prevents large-scale rework: Four upgrade stages with verifiable conditions, upholding the principle that "structure precedes scale".

User asset safety: Source directory is read-only with copy-only operations; cleanup moves files to .trash instead of deleting them.

Verifiable: Provides minimum standards and critical checkpoints for "system is functional", avoiding empty-shell deliveries.

Zero network, zero credentials: The 10 scripts rely solely on Node standard libraries fs/path and can run offline.

Best for: Suitable for users who have accumulated substantial local content assets (dozens of drafts, posts, course notes, or case studies) and want to convert "inventory" into a reusable engineering system: A typical workflow includes auditing volume and boundaries → initializing project → copying materials and indexing → extracting 15+ content units via samples → generating topic maps and assembly drafts → recombining real units into new topics; integrates with Obsidian [[filename]] links to generate node-relation graph views.

Not for: Not suitable for fragmented materials or single-article polishing (will be blocked at the audit gate with a downgrade path); not suitable for users expecting full repository processing in one shot (the skill only guarantees bringing it to a "functional state", with separate gates for full processing); not suitable for teams requiring cloud collaboration or multi-user concurrency (purely local Markdown + Node scripts); not suitable for publishing raw extraction outputs directly—both docs/acceptance.md and quickstart mandate human review.

Install Agent-Ready

① Verified Mirror 2026-09-15

Option A · Direct DownloadDownload dbs-content-system.tar.gz

sha256: b0d6fde8b85e94cf…

Option B · Copy Guide for Agent

Agents automatically download and verify mirrors from this JSON guide.

② Upstream GitHub

GitHub Repo ↗

Pinned to commit 8b8e33f1ec.