Guide · Agent skill
How to use the Feature Risk Assessment agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The feature-risk-assessment skill is an agent skill for in-house product legal work. When a launch review finds a risk that needs more than a single checklist line, it guides the agent to write a standalone decision document covering what could go wrong, how likely and severe each risk is, what already mitigates it, which regulators might care, and a recommended option. This guide explains what the skill does, how to install it, how to trigger it, and what it needs to run.
Feature Risk Assessmentanthropics/claude-for-legal · Apache-2.0 Download the skill Quick start
- 1Download and unzip. Download the skill ZIP from this page and unzip it. Place the folder in ~/.claude/skills/ for personal use or in the project's .claude/skills/ folder.
- 2Or upload in the Claude apps. If you use the Claude apps, go to Settings → Capabilities → Skills and upload the ZIP.
- 3Set up your practice profile. Make sure a practice-level CLAUDE.md exists at ~/.claude/plugins/config/claude-for-legal/product-legal/. The skill uses it for risk calibration.
- 4Ask for a deep dive. Describe the escalated feature with a request such as 'risk assessment for [feature]' or 'what could go wrong with [feature]'.
- 5Review the risks and options. Check each risk's likelihood, severity, mitigations and gaps, and compare the options in the table.
- 6Verify citations. Check every citation in the Regulatory landscape and Precedent sections against a primary source before the document goes to a decisionmaker.
What the skill makes the agent do
The skill turns a single escalated risk into a structured decision document. It is meant for the small share of launches that need this depth of scrutiny. The source says most launches do not, and a standard launch review is enough for them.
The document follows a fixed structure: a description of what is being assessed, a set of risks (aim for two to five), an optional regulatory landscape, optional precedent, two or three options compared in a table, and a recommendation. Each risk is scored for likelihood and severity with a stated reason rather than a bare rating.
- Describes the feature and why it was escalated
- Lists each risk with scenario, who is affected, likelihood, severity, existing mitigations, gaps and residual risk
- Compares realistic options and recommends one
Install the skill from the ZIP or the Claude apps
Download the skill ZIP from this directory page and unzip it. For a personal setup, place the unzipped skill folder in ~/.claude/skills/. For a single project, place it in that project's .claude/skills/ folder.
If you use the Claude apps instead, upload the ZIP under Settings → Capabilities → Skills.
- Personal use: unzip into ~/.claude/skills/
- Project use: unzip into the project's .claude/skills/
- Claude apps: Settings → Capabilities → Skills, then upload
Trigger the skill with natural requests
You do not need a special command. The skill's description lists phrases that should start it. Say something close to these when you are working on an escalated risk.
The skill also runs when a launch review flags a novel issue, so you can ask for it directly after a review finds something outside the standard checklist.
- "Deep dive on this risk"
- "Risk assessment for [feature]"
- "What could go wrong with [feature]"
- "Launch review flagged this as novel, run a risk assessment"
What you need before running it
The skill needs Claude Code or a compatible agent with access to the claude-for-legal product-legal plugin configuration files. It also reads a practice-level CLAUDE.md profile at ~/.claude/plugins/config/claude-for-legal/product-legal/. The source says this profile is referenced but not included, so you must create or obtain it yourself. The skill uses it for risk calibration, so the assessment reflects your company's regulatory posture and risk appetite rather than a generic view.
A legal research tool such as Westlaw or CourtListener is optional. It helps with verifying citations.
- Claude Code or a compatible agent with plugin config access
- Practice-level CLAUDE.md at ~/.claude/plugins/config/claude-for-legal/product-legal/
- Optional: Westlaw or CourtListener for citation checks
Safety notes: scripts, network and citations
The skill does not run scripts and does not use the network, according to its listing. Its main safety concern is the content it produces. The source states that citations in the Regulatory landscape and Precedent sections are AI-generated and must be checked against primary sources before a decision document reaches a decisionmaker. A risk assessment built on a fabricated enforcement action is worse than none.
If a research query returns few results, the skill is designed to report what it found and stop, and to ask you how to proceed rather than silently filling gaps from web search or model memory. Every citation is tagged with its source so you can see what needs checking.
- No scripts run and no network use
- Verify every citation against a primary source before sharing
- Citations are tagged with their origin, such as [Westlaw] or [model knowledge — verify]
Tips and related skills
If the risk involves an AI feature, the source suggests running the AI governance AIA skill in parallel or right after. The feature risk assessment frames the decision, and the AIA is the governance record. If the feature involves new data collection or processing, run the privacy PIA skill. The two documents will overlap, so flag the overlap to avoid duplicated work.
If the document will be shared outside the privileged group, the source says to drop the work-product header only for that external copy and keep the privileged original in the matter file.
- Use it only when the launch review is not enough
- Flag overlap with the PIA and AIA so work is not duplicated
- Keep the privileged original when making an external copy
/privacy-legal:pia-generation [feature]What people use it for
- An in-house team assessing a novel AI feature that the launch review did not have a checklist line for
- Product counsel preparing a decision document for a general counsel or leadership on a children's product feature
- A team flagged for a biometric or health feature that needs options compared before launch
Frequently asked questions
How do I install the feature-risk-assessment skill?
Download the skill 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 the ZIP under Settings → Capabilities → Skills.
What do I need to run this skill?
You need Claude Code or a compatible agent with access to the claude-for-legal product-legal plugin configuration files. You also need a practice-level CLAUDE.md at ~/.claude/plugins/config/claude-for-legal/product-legal/, which the source says is not included. A legal research tool such as Westlaw or CourtListener is optional.
Is this skill free to use?
The listing states the licence is Apache-2.0. Check the licence text in the download for the terms that apply to you. This page does not list a price.
Does the skill give a legal opinion?
No. The skill produces a decision document for a lawyer to review. Its citations are AI-generated and must be verified against primary sources before use, and it is not a finished compliance sign-off.
Does the skill run scripts or use the internet?
According to its listing, it runs no scripts and uses no network. It can use a legal research tool if you have one configured, and it asks before filling gaps from web search.
How is this different from a standard launch review?
A launch review covers most launches with a checklist. This skill is for the smaller set of cases where one risk needs a full written analysis with options and a recommendation. The skill itself says most launches do not need it.
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 in-house legal teams and product counsel who need a structured decision document for a novel or high-scrutiny feature, such as an AI, children's, biometric or health feature. It is a poor fit for routine launches, for anyone who needs a verified legal opinion, or for people outside legal who expect a finished compliance sign-off.
Download the skill