Guide · Agent skill
How to use the company-brainify agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
company-brainify is an agent skill that tells an AI agent how to turn a personal knowledge base, called a brain, into a version that is safe to share with a team. It searches the brain for internal ratings, compensation, performance assessments, retention talk, management strategy, political dynamics and personal details, then removes or edits them while keeping professional identity, projects, career history and expertise. After this guide you will know what the skill does, how to install it, how to ask for it in plain language, and what to check before you trust the result.
Company Brain Sanitizergarrytan/gbrain · MIT Download the skill Quick start
- 1Download the ZIP. Get the company-brainify ZIP from this skill's page in the directory.
- 2Install the skill. Unzip it into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder, or upload it under Settings, Capabilities, Skills in the Claude apps.
- 3Check your brain setup. Confirm gbrain is installed and sync.repo_path points to your brain repository.
- 4Make a request. Ask in plain language, for example to sanitize the brain before sharing it with the team.
- 5Test on a few files. Let the agent sanitize 3 to 5 files, then read the results yourself before going further.
- 6Verify with retrieval and back-links. Run the searches and check-backlinks step so no sensitive content or dangling links remain.
- 7Review history separately. Decide whether to purge git history, knowing it is destructive and can break collaborators' copies.
What the skill makes the agent do
An agent skill is a folder with a SKILL.md file that gives an AI agent specific instructions for one kind of task. This one covers sanitizing a personal brain of notes, people pages, meeting notes and daily logs so the team can use it.
The skill works across the full scan scope, not just the people folder. It checks people/, meetings/, daily/, companies/, projects/ and analysis/, and it also looks at takes and facts stored in the brain, plus back-links after edits. It can also run as a report-only audit of a shared brain that already exists.
- Removes internal scores and ratings, compensation, performance assessments, departure or retention talk, management strategy, political dynamics and personal details.
- Keeps name, role, work email, projects, career history, professional beliefs, work timeline and expertise.
- Deletes fact rows that carry sensitive claims instead of only expiring them.
- Checks git history for older versions that still contain sensitive content, and can purge that history behind a confirmation step.
Install company-brainify in Claude
Download the skill as a ZIP from the skill's page in this directory. Then choose one of the two install routes below. Both use the same skill files.
For a personal install, unzip the file into ~/.claude/skills/. For a single project, unzip it into the project's .claude/skills/ folder. You can also upload the ZIP in the Claude apps under Settings, then Capabilities, then Skills.
The skill's files are SKILL.md only, so after unzipping you should see a company-brainify folder containing SKILL.md.
- Personal: unzip into ~/.claude/skills/
- Project: unzip into .claude/skills/ inside the project
- Claude apps: Settings → Capabilities → Skills, then upload the ZIP
Ask for it with natural requests
You do not need a special command. The skill is triggered when your request matches its trigger phrases. Describe what you want in plain language and mention the brain or the team.
The trigger phrases include 'company brain', 'team brain', 'brainify', 'sanitize the brain', 'share my brain with the team', 'strip sensitive data from the brain', 'scrub employee data', 'audit the shared brain' and 'make the brain safe to share'.
- 'Sanitize the brain so I can share it with my team.'
- 'Audit the shared brain for anything that should not be there.'
- 'Make the brain safe to share, and show me what you would remove first.'
- 'Scrub employee data from the team brain.'
Run a sanitization pass safely
The skill's procedure starts by setting the brain's location from your gbrain configuration, then moves to the repository folder. It then runs retrieval searches to find sensitive content, and it writes the matching page paths into a scope file. Structural keyword searches add to that same list.
The skill says to sanitize 3 to 5 files first, read the output yourself, and only then ramp up. A bad bulk pass can look finished while still leaking, so manual review is required.
- Set the location: BRAIN="$(gbrain config get sync.repo_path)" then cd into it.
- Discovery uses gbrain query with a limit, gbrain takes search and gbrain recall --grep.
- Grep patterns are triage tools. The model makes the judgment on each file.
- After edits, gbrain check-backlinks check confirms no page still points at removed sections.
BRAIN="$(gbrain config get sync.repo_path)"
cd "$BRAIN"
gbrain query "compensation, equity, or salary discussions about team members" --limit 50
gbrain takes search "performance" --limit 50
gbrain recall --grep "salary"What it needs and what it touches
The skill needs the gbrain command-line tool with a configured brain repository. It also needs a Git repository for the brain if you want the history purge step, and access to gbrain query, takes search, recall and check-backlinks.
The skill runs scripts. It does not use the network, according to its listing. It edits pages in place inside the existing folders and writes a date-keyed deletion log under daily/.
- Requires gbrain with sync.repo_path configured.
- Requires a Git repository for history purge.
- Requires access to retrieval commands: gbrain query, takes search, recall, check-backlinks.
- Runs scripts: yes. Uses network: no.
Safety notes before you run it
Purging git history rewrites the repository, which is destructive and can break collaborators' copies. Only do it on a repository that others do not depend on, and only after the confirmation step.
Do not use this skill to import or connect an existing company repository. The skill says that is a different workflow, and stripping metadata is not needed there. For that case, start with gbrain sources inspect <path> --profile company-brain and follow the company brain ingestion guide.
Expect manual review. Judgment calls are made per file, so the output is not a guaranteed redaction.
- Back up the brain before a bulk pass.
- Read every changed file, not just the list.
- Do not assume a clean working tree means clean history.
What people use it for
- Founders or executives preparing a shared brain for their team without HR-sensitive notes.
- Teams auditing an existing shared brain for content that should not be there.
- Maintainers of the gbrain project who need a repeatable sanitization workflow.
Frequently asked questions
How do I install the company-brainify skill?
Download the ZIP from this skill's page. Unzip it into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder. In the Claude apps you can instead upload the ZIP under Settings, Capabilities, Skills.
What do I need to run company-brainify?
You need the gbrain command-line tool with a configured brain repository at sync.repo_path, access to retrieval commands like gbrain query and takes search, and a Git repository if you want to purge history.
Is company-brainify free?
The skill is listed under the MIT licence. This page does not state any price, so check the source listing for current terms.
Does it guarantee a fully automatic redaction?
No. The skill makes judgment calls per file and warns that bulk passes can look finished while still leaking. Manual review is required.
Can I use it to import an existing company repository?
Not as a sanitization job. The skill says connecting or importing an existing repository is a different workflow. Use gbrain sources inspect with the company-brain profile and follow the ingestion guide.
What should I do if sensitive content is still found after a pass?
Re-run the retrieval searches and the grep checks on the flagged pages, then use the check-backlinks step to find pages that still point at removed sections. Review those files manually.
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 company-brainify if you are a founder, executive or team maintainer who needs to share a personal brain with colleagues and wants a structured, repeatable review process. Pick something else if you only need to import an existing repository, if you want guaranteed automatic redaction, or if you cannot safely rewrite git history.
Download the skill