How to use the Cold-Start Practice Interview agent skill

What the cold-start-interview skill does

The skill walks a lawyer through a first-time interview about their IP practice. Topics include enforcement thresholds, approval chains, brand watch settings and open-source rules. It can also read practice documents you share, such as a portfolio list or IP management export, brand guidelines, cease-and-desist templates, an enforcement playbook and an open-source policy.

The output is a profile file written to your plugin config folder: ~/.claude/plugins/config/claude-for-legal/ip-legal/CLAUDE.md. The skill is meant to be the first thing you run on a fresh install. Other skills in the same plugin read this profile so their output matches your practice's positions.

• Asks about your practice and extracts stated positions from your documents
• Notes gaps between your stated positions and what your templates or playbooks require
• Writes the profile in plain English that you can edit by hand
• Can re-check which external tools are connected, without re-running the interview

What you need before you install

The skill depends on the claude-for-legal ip-legal plugin. It is not meant to run on its own, and it relies on that plugin's template and other skills. You also need Claude Code with that plugin installed and access to the local ~/.claude/plugins config folder, because the skill writes files there.

• Claude Code with the claude-for-legal ip-legal plugin installed
• Access to the local ~/.claude/plugins config folder for writing files
• Optional: an IP management system, patent or legal research tool, document storage, or Slack connection (MCP) for integration checks

Install the skill from a ZIP or the Claude apps

Download the cold-start-interview ZIP from this page. Then choose one of two install routes. For a personal install available in all your projects, unzip it 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, open Settings, go to Capabilities, then Skills, and upload the ZIP there.

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

Run the interview and trigger the skill

Run the skill with its slash command. The source gives these examples: running it with no flag starts the first-time interview, --redo re-runs it on an already configured plugin, and --check-integrations re-probes integrations only.

You can also ask in plain language. The skill description lists these situations as triggers: first install when the profile is missing or still has placeholders, re-onboarding after your enforcement posture changes, and re-checking integrations after you connect or disconnect an MCP. The source's examples of natural requests include asking to redo the interview or saying your enforcement posture has changed.

• /ip-legal:cold-start-interview (first-time setup)
• /ip-legal:cold-start-interview --redo (re-interview an already configured plugin)
• /ip-legal:cold-start-interview --check-integrations (re-probe connected tools only)
• Natural request: 'let's redo the interview' or 'my enforcement posture changed'

What happens during the interview

The skill first checks whether your profile exists and whether it has placeholder text. If the profile is missing or still a template, it starts the interview. If the profile is already filled in and you did not pass --redo, it stops and asks before doing anything else.

It asks for your practice documents and accepts file paths, Google Drive links or IP management record IDs. It then writes the profile, and if you shared a portfolio export or have IP management access, it also writes a portfolio register file. At the end it shows you a summary and asks what it got wrong.

• Writes ~/.claude/plugins/config/claude-for-legal/ip-legal/CLAUDE.md
• Writes ~/.claude/plugins/config/claude-for-legal/ip-legal/portfolio.yaml if a portfolio was shared, otherwise leaves a placeholder pointer
• With --redo, shows a diff before overwriting the existing profile
• If a profile exists only in the old cache path, copies it to the config path first and shows what was migrated

Safety notes: scripts, network and overwrites

According to the skill listing, this skill runs no scripts and uses no network access. Its work is reading your documents and writing the profile files. The main risk is overwriting your existing setup, so re-runs with --redo show a diff first.

The integration check only reports a connection as working when a tool call actually succeeds. Connectors that are declared but not tested are marked as untested, so a declaration in configuration does not count as proof that a tool works.

• No scripts are run and no network use is listed
• Re-running with --redo shows a diff before overwriting
• Profile is a plain-language document you can edit directly
• Integration status is only ✓ after a successful tool call