Guide · Agent skill
How to use the Deep-Company Series Writer agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The deep-company-series skill is an instruction file that tells an AI agent how to write an eight-part, long-form deep dive on one public company, roughly 120,000 words in total, aimed at public readers. Its main focus is revision: a fact-check checklist that catches false precision, absolute wording and numbers that disagree across articles. This guide explains what an agent skill is, how to install this one, how to trigger it, what it needs, and where it should not be used.
Deep-Company Series WriterHKUDS/Vibe-Trading · MIT Download the skill Quick start
- 1Download the ZIP. Get the deep-company-series ZIP from the directory page.
- 2Unzip into a skills folder. Place the folder in ~/.claude/skills/ for personal use, or in a project's .claude/skills/ folder.
- 3Or upload in the Claude apps. Go to Settings → Capabilities → Skills and upload the skill.
- 4Confirm required tools exist. Check that your environment provides financial statement, web research and file-writing tools.
- 5Make a series request. Name the company and ask for an eight-part deep dive for public readers.
- 6Review each part. Read each part as it is written and give feedback before the next one starts.
- 7Run the consistency scan. After all parts are drafted, ask for the cross-article check of numbers and absolute language.
What is an agent skill, and what does this one do?
An agent skill is a folder containing a SKILL.md file with instructions that an AI agent loads when a request matches the skill's description. The skill does not add new software; it shapes how the agent plans and writes its work.
This skill produces a series of about eight articles on a single company. The parts run from an opening reset of common assumptions, through the company's moat, profit engines, hidden assets, and the impact of an era variable such as AI, then financials, management, and a valuation and decision framework. Each part is written to stand alone while sharing one valuation, management and price framework across the series.
- Part 01: cognitive reset, breaking common illusions about the company
- Parts 02-03: moat and the most profitable business
- Part 04: hidden assets on the balance sheet
- Part 05: era variable, decomposed by business
- Part 06: financials read in a Buffett-style way
- Part 07: management and capital allocation
- Part 08: valuation, three scenarios, red lines and position framework
Install the skill from the ZIP file
Download the ZIP for deep-company-series from the directory page. Unzip it into your personal skills folder, ~/.claude/skills/, or into a project's .claude/skills/ folder if you want the skill available only for that project.
If you use the Claude apps instead, upload the skill under Settings → Capabilities → Skills. The source does not describe other installation methods, so do not assume a package manager command exists for this skill.
- Personal install: unzip into ~/.claude/skills/
- Project install: unzip into the project's .claude/skills/
- Claude apps: Settings → Capabilities → Skills, then upload
How to trigger the skill with a natural request
You do not need a special command. Describe what you want in plain language, naming the company and the format. The skill is meant for requests for textbook-level depth on one company for public publishing, delivered as a multi-part series.
Make clear that you want a series, not a single report. Requests for a single research report, earnings note or sector study should go to other skills, such as investment-research or earnings-review.
- "Write an 8-part deep-dive series on [company] for public readers."
- "Help me outline a long-form series on [company]'s moat and management."
- "Fact-check my multi-part company series for inconsistent numbers and absolute claims."
- "Draft part 08 of my series on [company] with a three-scenario DCF and red lines."
What the skill needs to run
The skill assumes the host environment provides tools. The source names financial data tools such as get_financial_statements, web research via get_research_reports or web_search, an optional run_swarm tool for an internal multi-agent draft, and a file-writing tool to save articles to a reports folder. If these tools are missing, the workflow cannot run as written.
The skill makes network calls, because it relies on live financial and web tools. It does not state that it runs scripts.
- Financial data tool such as get_financial_statements (the source says the host must provide it)
- Web search or research report access such as get_research_reports or web_search
- Optional run_swarm for an internal draft
- A file-writing tool to save parts to a reports folder
- Network access: yes. Scripts: none listed
Safety notes on data, network and claims
The skill cannot guarantee the accuracy of its sources, since it depends on whatever financial and web tools the agent has. Verify important numbers against filings yourself before publishing. The skill also tells the agent to confirm the core theses with you before writing parts 01 and 02, and to wait for your review between parts rather than publishing immediately.
The output is public-facing commentary and a valuation framework. It is not personal investment advice or a trading signal, and it should not be read as one.
- Confirm the core theses with the user before writing begins
- Review each part before publishing; the skill does not publish automatically
- Verify facts with get_financial_statements or web_search when you dispute a number
- Do not use the output as a personal buy or sell decision
Tips for better results
The cross-article consistency scan is the most distinctive part of the skill. Run it after all parts are drafted, so market cap, net income, holding percentages and P/E values are checked across the whole series. The skill also asks the agent to define terms such as SBC and Non-IFRS on first use and to label which financial measure each number uses.
Expect the skill to remove weighted-probability expected returns and to refuse to attribute a competitor's failure to a single cause. If you want those, the skill is working against your request.
- Give the agent the company name and your target readers at the start
- Ask for a consistency scan after all eight parts are drafted
- Keep a list of sources you accept, so third-party user numbers can be marked with their source
- Check that cross-references such as 'see part 06' point to real sections
What people use it for
- Drafting an eight-part public series on one listed company for a newsletter or publication
- Checking a finished multi-part company series for inconsistent market cap or net income figures
- Outlining a long-form company research project with a shared valuation framework
- Removing absolute or hype language from an existing finance long-form before publication
Frequently asked questions
How do I install the deep-company-series skill?
Download the ZIP from the directory page and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. In the Claude apps, you can instead upload it under Settings → Capabilities → Skills.
Is the deep-company-series skill free?
The source material is licensed under MIT. The source does not state pricing for the skill itself, so check the directory listing for any cost details.
What do I need to run this skill?
Your environment must provide financial statement and web research tools, such as get_financial_statements and get_research_reports or web_search, plus a file-writing tool. run_swarm is optional. The skill uses the network but the source lists no scripts.
Can I use it for a single earnings note or research report?
No. The skill is for a multi-part series of about eight articles. Single research reports, earnings notes and sector studies should use the investment-research or earnings-review skills instead.
Will this skill give me investment advice or buy and sell signals?
No. It produces public-facing commentary and a valuation framework. It is not meant for personal investment decisions or trading signals, and the output should not be read as advice.
How does this compare with other company research skills?
This skill focuses on long, publication-style series and on revision checks for consistency and overconfident claims. For a short report or earnings review, a different skill fits better. Compare the depth and format you need before choosing.
Why does the skill refuse probability-weighted forecasts?
The skill treats weighted expectations built from subjective probabilities as false precision. It asks the agent to list scenarios, triggers and direction instead, and to use high and low ranges rather than promised figures.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Choose this skill if you are writing or checking a long, multi-part public deep dive on one company and want strict checks on numbers and wording. Pick a different skill for single reports, earnings notes, sector studies, personal investment decisions or any task needing guaranteed real-time data.
Download the skill