Business judgment, content production and lasting records
dbskill
Choose a Method for Real Business Work
dontbesilent turns business and content methods into an Agent Skills collection. Give the dispatcher a real task, then work toward a diagnosis, plan, draft or file. When several capabilities are needed, it explains the division of work and continues in the same conversation.

Project demo
View example sourceHosts listed by the author
- Claude Code
- Codex
- WorkBuddy
- Doubao
- GitHub stars
- 10,946
- Version
- 3.1.0
- Checked
- Categories
- Content & Knowledge
Useful when business facts, content material and the next action belong in one conversation
Begin with a defined goal: investigate a pricing problem, plan a chosen topic or review a complete draft. Judge the result by the facts it uses, the output it delivers and whether the next action can be tested.
Pick a task and get started
These examples use the full Claude Code plugin. Run the two terminal commands, confirm /dbs:dbs is available in chat, then choose a task and replace the example material. Other installation routes use different invocation names, explained below.
Use the dispatcher for a first task. Include the product, price, buyer, acquisition channel and observed feedback so it can identify what needs checking. Replace this editorial cake-studio scenario with your own business.
Install the full Claude Code plugin
Run both lines in a terminal to install the full dontbesilent Claude Code plugin. Use /dbs:dbs for the entry and names such as /dbs:dbs-diagnosis for members; the examples match this route. The page also explains the Skills CLI alternative and its shorter names.
Read the author’s setup guideclaude plugin marketplace add dontbesilent2025/dbskill claude plugin install dbs@dontbesilent-skillsMatch the chat command to the installation route
The plugin entry is /dbs:dbs; members use names such as /dbs:dbs-diagnosis. A Skills CLI installation typically uses /dbs or the invocation shown by the host. If commands are missing, check that installation succeeded and targeted the agent you are using. Keep the installation route and invocation consistent.
Read the author’s installation and invocation notesGive your AI this first task
These editorial examples follow the author’s documentation. The cake studio is an illustrative scenario, not an author case study or a site-tested result. Replace the business data and attach your files. Commands match the Claude Code plugin route shown here.
/dbs:dbs I run an appointment-only cake studio. Our main product is a CNY 288 birthday cake for local families. Customers arrive through Xiaohongshu and referrals; we make the cake and deliver locally. Last month revenue was about CNY 30,000. Of the last 20 new inquiries, eight people said the price was high and four stopped replying. I have not yet recorded their budgets or whether they eventually bought. I want to decide whether to lower the price. Check the question and available evidence, name the Skill you actually use and explain why, then begin in this conversation. Separate facts, inferences and missing information. Give a conditional judgment and one action I can take tomorrow. Do not treat “too expensive” as proof of a pricing error or invent sales records.
Check what you receive
- An explanation of the selected skill and its purpose, followed by analysis or a question about a fact that could change the answer.
- A judgment, supporting reasons and a practical validation step. A full business-model assessment is a separate request, not the default for every question.
- Check that prices, customer details and sales figures come from your input, that experience-based judgments are identified, and that the proposed action can be tested in the business.
More about this skill
dbskill is a Chinese-first collection by dontbesilent for people operating a business and producing content. Its examples include courses and consulting as well as shops, goods and services. What matters is the actual product, customer, transaction and delivery information you provide. The methods have a distinct point of view: separate supplied facts, analytical inferences and the author’s experience-based positions.
The /dbs dispatcher recovers the current goal, material and feedback, then works backward from the desired deliverable. It can answer a simple question directly, use one skill, or arrange a necessary sequence, separate analyses, comparison or revision. It focuses on one current goal; it is not restricted to selecting one skill per turn and does not run the entire collection by default.
Content members have different deliverables. dbs-content plans a chosen topic; dbs-resonate locates resonance problems in an existing draft; dbs-title-cover-intro supplies titles, cover text and openings; dbs-plain-writing can produce factual explanatory copy. Decide whether you need a judgment, revision instructions or usable text before choosing between adjacent methods.
You can add facts and feedback within the same conversation. Continuing across conversations requires an explicit save. dbs-save, dbs-restore and dbs-report use local Markdown to preserve conclusions, rejected directions and open hypotheses. The repository’s public posts and knowledge atoms are optional reference material: installing the skills does not download the whole corpus or automatically build a private knowledge base.
Sources and verification
Sources checked
Checked against the author’s README, dispatcher and representative skills. Installation commands and task outputs were compared with the documentation. We did not install the plugin, run business diagnoses, call external services or test generated results.