Skip to content
Get packs

Guide · Agent skill

How to use the Privacy Plugin Cold-Start Interview agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The cold-start-interview skill is a one-time setup interview for the privacy-legal plugin in the claude-for-legal project. It reads your privacy policy, a DPA template and one reference privacy impact assessment (PIA), then writes a plain-text CLAUDE.md file that records how your privacy team works so other privacy skills read the same understanding. This guide explains what the skill does, how to install it, how to start it with natural requests, and what to check before you run it.

Privacy Plugin Cold-Start Interviewanthropics/claude-for-legal · Apache-2.0

Quick start

  1. 1Get the plugin and the skill. Install the claude-for-legal privacy-legal plugin first. Then download the cold-start-interview ZIP from the Privacy Plugin page on this directory.
  2. 2Unzip into the right folder. Unzip the skill into ~/.claude/skills/ for a personal install, or into .claude/skills/ inside a project. In the Claude apps, upload the ZIP under Settings → Capabilities → Skills.
  3. 3Gather your seed documents. Have your privacy policy (URL or file), your DPA template and one reference PIA ready before you start.
  4. 4Start the interview. Run /privacy-legal:cold-start-interview or say 'set up the privacy plugin'. Choose quick (about 2 minutes) or full (about 15 minutes) when prompted.
  5. 5Answer the questions and review the summary. Answer the questions. The skill writes CLAUDE.md to the config path and shows you a summary before offering a first task.
  6. 6Check integrations later if needed. After connecting or disconnecting a document storage, Slack or scheduled-task connector, run /privacy-legal:cold-start-interview --check-integrations.

What does the cold-start-interview skill make the agent do?

The skill runs the first-time configuration for the privacy-legal plugin. It asks you questions, reads three seed documents (your privacy policy, your DPA template and one reference PIA), and extracts your policy commitments, your DPA positions and the structure of your PIA.

It then writes the result to ~/.claude/plugins/config/claude-for-legal/privacy-legal/CLAUDE.md. Every other privacy skill reads that file, so the output of later tasks reflects how your team works rather than a generic template. The skill can also re-check which integrations are working without re-running the full interview.

  • Quick path: about 2 minutes, covering your role, which side of a DPA you sit on (processor, controller or both) and primary jurisdictions, with defaults for everything else.
  • Full path: about 15 minutes, adding DPA playbook positions, PIA template structure from a reference PIA, your full regulatory footprint and processing-activity seeds.
  • Shared company profile: if ~/.claude/plugins/config/claude-for-legal/company-profile.md exists, the skill reads it and skips company questions that other legal plugins already cover.

How do you install the cold-start-interview skill?

Download the skill ZIP from the Privacy Plugin page in this directory. Then choose one of two install locations.

For a personal install that works across your projects, unzip the file into ~/.claude/skills/. For a project-only install, unzip it into the .claude/skills/ folder inside that project.

If you use the Claude apps instead of Claude Code, upload the ZIP under Settings → Capabilities → Skills.

  • Personal install: ~/.claude/skills/
  • Project install: .claude/skills/ inside the project folder
  • Claude apps: Settings → Capabilities → Skills, then upload the ZIP

What do you need before you run it?

The skill configures a specific plugin, so the privacy-legal plugin must already be installed. You also need access to your privacy policy (as a URL or a file), your DPA template and one reference PIA. The skill reads all three during setup.

The skill needs read and write access to ~/.claude/plugins/config and ~/.claude/plugins/cache. Connectors for document storage, Slack and scheduled tasks are optional. The skill's own notes say these connectors were not tested by this file.

  • The claude-for-legal privacy-legal plugin installed
  • Your privacy policy (URL or file), DPA template and one reference PIA
  • Read/write access to ~/.claude/plugins/config and ~/.claude/plugins/cache
  • Optional: MCP connectors for document storage, Slack and scheduled tasks
  • Optional: a shared company-profile.md and references/company-profile-template.md in the plugin root

How do you trigger the skill with natural requests?

You can start the interview by typing the slash command, or by asking in plain language. The skill is designed to respond to requests such as setting up the privacy plugin, asking to be onboarded, configuring privacy, or re-running the interview.

The skill also runs when the configuration file is missing or still contains placeholders. To re-check only the integrations, ask for that directly or use the flag shown below.

  • 'Set up the privacy plugin.'
  • 'Onboard me.'
  • 'Configure privacy.'
  • 'Re-run the privacy interview.'
  • 'Re-check my integrations.'
/privacy-legal:cold-start-interview /privacy-legal:cold-start-interview --check-integrations

What do the flags do?

The --redo flag re-runs the interview. If a populated CLAUDE.md already exists, the skill asks you to confirm before it overwrites that file.

The --check-integrations flag re-runs only the integration availability check for document storage, Slack and scheduled tasks, and updates the Available integrations section. It does not re-interview you. Use it after you connect or disconnect an MCP connector.

  • --redo: re-run the full interview (confirmation needed before overwriting)
  • --check-integrations: re-probe integrations only

Is it safe to run, and what does it touch?

The skill does not run scripts. It does use the network, because it reads your policy if you give a URL. It also reads the DPA template and reference PIA you supply, and it processes those documents as part of setup. If you do not want the agent to read and summarise those documents, do not run this skill.

The skill writes files that persist in your home directory, and other skills rely on them. The skill only reports a connector as working (✓) after a tool call to it has succeeded. Connectors that are configured but untested are marked ⚪, with a note on how to confirm them.

  • Runs scripts: no
  • Uses network: yes (for example, fetching a policy URL)
  • Writes: ~/.claude/plugins/config/claude-for-legal/privacy-legal/CLAUDE.md
  • Migration: copies a populated CLAUDE.md from the old cache path into the config path if needed, and shows you what moved
  • Integration status is reported as ✓ only after a successful test call, never from .mcp.json declarations alone

What people use it for

  • A privacy lead onboarding the privacy-legal plugin so later DPA reviews follow the team's own playbook positions.
  • A privacy team re-running setup after its privacy policy, DPA template or PIA format changes, using --redo.
  • A team that connected Slack or a document storage connector and wants the plugin to detect it, using --check-integrations.
  • A privacy professional who wants PIA drafts to follow the house style taken from a reference PIA.

Frequently asked questions

Is the cold-start-interview skill free to use?

The source file does not state a price or licence terms for the skill beyond its Apache-2.0 licence, which is listed on this page. Check the licence file in the repository for the terms that apply to your use.

What do I need to install before the cold-start-interview skill?

You need the claude-for-legal privacy-legal plugin installed, plus access to your privacy policy, your DPA template and one reference PIA. The skill also needs read and write access to ~/.claude/plugins/config and ~/.claude/plugins/cache.

Where do I put the cold-start-interview skill files?

Unzip the downloaded ZIP into ~/.claude/skills/ for a personal install, or into .claude/skills/ inside a project for a project-only install. In the Claude apps, upload the ZIP under Settings → Capabilities → Skills.

Does the skill run scripts or use the internet?

It does not run scripts. It does use the network, for example when your privacy policy is given as a URL.

Can I re-run the setup or only check my integrations?

Yes. Use --redo to re-run the full interview, which asks for confirmation before overwriting a populated CLAUDE.md. Use --check-integrations to re-probe document storage, Slack and scheduled-task connectors without re-interviewing.

Who should not use this skill?

Anyone who only wants general privacy advice should not use it, because it is built to configure a specific plugin. It is also a poor fit if you do not want the agent to read your policy, DPA template and PIA, or if you use a plugin other than privacy-legal.

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

This skill suits privacy or legal team members who are setting up the claude-for-legal privacy-legal plugin for the first time, or who need to re-run setup after their policy, DPA template or PIA format changes. If you only want general privacy advice, or you do not want the agent to read your policy and PIA documents, pick a different tool.

Download the skill