Guide · Agent skill
How to use the Company Context Engine agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The Company Context Engine is an agent skill that gives your AI assistant a persistent picture of your company. It reads a profile file on your computer, warns you when that profile is out of date, and follows rules about what company information can leave your machine. This guide is for founders and executives who use several C-suite advisor skills and want advice fitted to their situation; by the end you will know what the skill does, how to install it, how to trigger it and where its limits are.
Company Context Enginealirezarezvani/claude-skills · MIT Download the skill Quick start
- 1Download and unzip. Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use or the project's .claude/skills/ folder.
- 2Or upload in the Claude apps. In the Claude apps, open Settings, go to Capabilities, then Skills, and upload the skill.
- 3Build your company context. Run /cs-onboard setup from the wider repository to create ~/.claude/company-context.md.
- 4Check the file. Open the file and correct any inaccurate stage, team, challenge or financial details.
- 5Start an advisor session. Ask a normal question in a C-suite advisor skill and the context engine loads first.
- 6Refresh when prompted. If the file is 90 days or older, run /cs-onboard update or continue with the stale context.
- 7Review before sending externally. Before any web search or external service call, check that the anonymized content contains no sensitive figures or names.
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 agent such as Claude Code reads and follows when a task matches its description. This skill is the memory layer for the C-suite advisor skills in the same repository. Every advisor skill loads it first, so advice is based on your stage, team and priorities rather than generic assumptions.
At the start of a session the skill looks for ~/.claude/company-context.md. If the file is there, it reads the Last updated field and loads the key fields into working memory: company stage, founder archetype, current #1 challenge, runway, team size, unfair advantage and 12-month target.
- Reads ~/.claude/company-context.md at the start of a C-suite session
- Flags context older than 90 days and offers a refresh or lets you continue
- Notes new facts learned during a conversation and asks before saving them to the file
- Applies anonymization rules before any external API call or web search
How do you install the context-engine skill?
Download the ZIP for this skill from this page and unzip it. You can install it in one of two places. For personal use across all projects, put the unzipped folder in ~/.claude/skills/. For a single project, put it in that project's .claude/skills/ folder.
If you use the Claude apps rather than Claude Code, you can upload the skill instead. Open Settings, go to Capabilities, then Skills, and upload the skill there.
- Personal install: unzip into ~/.claude/skills/
- Project install: unzip into the project's .claude/skills/ folder
- Claude apps: Settings → Capabilities → Skills → upload
How do you create the company context file?
The skill does not write the company profile for you on first use. If no file exists, it asks you to run /cs-onboard setup. That command comes from the wider repository, not from this skill, so you need the other skills from the same repository installed to use it.
The file records items such as what your company does, your stage, your founder profile, your top challenge and your 12-month goal. The skill treats the profile as the source of truth, so the more accurate it is, the more useful the advice.
- Run /cs-onboard setup to build the file at ~/.claude/company-context.md
- Review the file yourself, since advice relies on what it says
- Keep the Last updated date current so the skill can judge freshness
/cs-onboard setupHow do you trigger the skill with natural requests?
You do not need a special command to use the skill. It is meant to run when you start any C-suite advisor session, when context looks stale or missing, or before you send company data to an external service. Ask your advisor a normal question and the skill loads its context first.
If the file is more than 90 days old, expect a prompt offering a quick refresh with /cs-onboard update, or the option to continue with the existing profile. If you continue, the agent notes that the context is stale and works with it.
- "Start a strategy session. Use my company context."
- "My context might be out of date, can you check before we plan hiring?"
- "Before you look up competitor pricing, make sure we don't share our revenue."
- "I just closed a new enterprise customer. Remember that for our next session."
What privacy rules does the skill apply?
The skill sorts information into two groups. Some items are never sent externally, including specific revenue or burn figures, customer names, employee names unless public, investor names unless public, specific runway months and the contents of your watch list. Other items are safe with anonymization, such as your stage label, team size range, industry vertical, challenge category and market position descriptor.
Before any external API call or web search, the skill follows references/anonymization-protocol.md. Numbers become ranges or stage-relative descriptors, names become roles, revenue becomes percentages or stage labels, and customers become Customer A, B and C. Local reads and writes of the context file and local reasoning are outside that protocol.
- Never sent: revenue or burn figures, customer names, investor names, runway months, watch list
- Safe with anonymization: stage label, team size range, industry, challenge category
- Numbers become ranges; names become roles; customers become Customer A, B, C
What should you check before relying on it?
The privacy rules are instructions the agent follows. They are not technical controls enforced by software, so review anything the agent plans to send outside your machine. The skill also cannot verify your financials; it relies on what you wrote in the profile file.
The skill runs no scripts and makes no network calls of its own. Its work happens inside the agent session, and it stores data only in the local context file. Keep that file private, because it can contain sensitive business information.
- Review outgoing content before any external search or service call
- Check that the financial and team details in the file are accurate
- Keep the file on your machine and treat it as sensitive
- The skill needs Claude Code or a compatible agent that reads SKILL.md files
What people use it for
- A founder preparing a board update gets advice that reflects their stage, team size and current top challenge.
- A small leadership team shares one reviewed company profile so each session starts from the same facts.
- A user asks for competitor research and the skill anonymizes revenue and customer details before the search is made.
- A CEO records a newly closed deal during a conversation and approves adding it to the file at session end.
Frequently asked questions
Is the context-engine skill free to use?
The skill is licensed under MIT, as stated in its SKILL.md. The source does not describe any pricing, so check the repository for the terms of any other tools you use with it.
What do I need to use the context-engine skill?
You need Claude Code or a compatible agent that reads SKILL.md files, a company context file at ~/.claude/company-context.md, and other C-suite advisor skills from the same repository to make the loaded context useful.
Where do I put the context-engine skill files?
Unzip the downloaded ZIP 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.
Does the context-engine skill send my company data to the internet?
The skill itself does not make network calls and runs no scripts. It applies anonymization rules before any external API call or web search, but those rules are instructions the agent follows rather than enforced technical controls, so review what is sent.
Why does the skill say my context is stale?
It checks the Last updated field in ~/.claude/company-context.md. If the date is 90 days or more in the past, it flags the context as stale and offers a refresh with /cs-onboard update.
Can the context-engine skill replace a financial analyst?
No. The skill relies on whatever the founder wrote into the profile file and does not provide verified financial analysis. Use it for tailored advisory conversations and check financial figures separately.
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
Use this skill if you regularly work with several C-suite advisor skills and want them to remember your stage, team and priorities without re-explaining them each session. Pick something else if you need guaranteed privacy controls, do not want a persistent file of sensitive business data on your machine, or need verified financial analysis.
Download the skill