Guide · Agent skill
How to use the Diligence Issue Extraction agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The Diligence Issue Extraction skill is an agent skill that tells an AI agent how to work through a deal's virtual data room (VDR), the online folders of documents used in M&A due diligence. It maps folders to a request list, filters for material documents using your firm's thresholds, checks documents against standard concerns, and writes findings in house memo format. This guide covers what the skill does, how to install it, how to trigger it, and what it needs before you rely on it.
Quick start
- 1Download the ZIP. Download the diligence-issue-extraction ZIP from this page. It contains SKILL.md.
- 2Unzip into a skills folder. Unzip it into ~/.claude/skills/ for all projects, or into a project's .claude/skills/ folder for one project.
- 3Or upload in the Claude apps. Alternatively, go to Settings → Capabilities → Skills in the Claude apps and upload the skill.
- 4Set up configuration files. Create CLAUDE.md and deals/[code]/deal-context.md under ~/.claude/plugins/config/claude-for-legal/corporate-legal/ with your categories, thresholds and memo format.
- 5Connect optional tools. If you have them, connect a VDR connector (Box, Intralinks or Datasite) and a legal research connector (such as Westlaw or CourtListener).
- 6Trigger the skill. Ask in plain language, for example "Review the data room" or "Extract issues from /02-Contracts".
- 7Review the output. Check the inventory and gaps, then review each finding and its tagged citations before relying on it.
What is an agent skill and what does this one do?
An agent skill is a folder containing a SKILL.md file that gives an AI agent specific instructions for a task. When a request matches the skill's description, the agent follows its workflow instead of improvising.
This skill, diligence-issue-extraction, is for a first-pass review of a large data room. It inventories VDR folders against a request list and notes gaps where the request list has no matching content. It applies a materiality filter so the agent does not review every document. For each document it reads, it checks standard concerns for that category, such as change-of-control clauses, assignment restrictions, IP ownership gaps, and litigation. Findings are written in the house memo format, and legal citations are tagged by source so reviewers can see what needs verification.
- Inventory VDR folders and flag request categories with no content
- Filter documents by materiality thresholds in your configuration
- Check material contracts for change of control, assignment, exclusivity, MFN, termination rights and unusual indemnities
- Check corporate, IP, employment and litigation documents against their standard extraction sets
- Tag each citation with its source, for example [Westlaw], [CourtListener] or [model knowledge — verify]
Install the skill
Download the diligence-issue-extraction ZIP from this page and unzip it. The package contains a single file, SKILL.md. You can install it in one of two places, depending on whether you want it available everywhere or only in one project.
For all your projects, unzip the folder into ~/.claude/skills/. For a single project, unzip it into that project's .claude/skills/ folder.
Alternatively, upload the skill in the Claude apps under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside the project folder
- Claude apps: Settings → Capabilities → Skills, then upload
~/.claude/skills/Set up the required configuration files
The skill depends on configuration files from the Claude for Legal corporate-legal plugin. It reads them before doing any work. Without them, the skill does not run correctly, so set them up first.
The source requires two files under ~/.claude/plugins/config/claude-for-legal/corporate-legal/: CLAUDE.md, which holds your diligence categories, materiality thresholds and issues memo format, and deals/[code]/deal-context.md, which holds deal-specific thresholds and the VDR location. If deal-context.md does not exist, the skill asks which deal you mean.
- Required: CLAUDE.md in the corporate-legal config folder
- Required: deals/[code]/deal-context.md for each deal
- Optional: a VDR MCP connector for Box, Intralinks or Datasite, which lets the skill pull the index directly
- Optional: a legal research connector such as Westlaw or CourtListener for citation checks
~/.claude/plugins/config/claude-for-legal/corporate-legal/How to trigger the skill with natural requests
You do not need to type a command. The skill is designed to activate when you use phrases like the ones below, or when you point the agent at VDR documents. The source also lists /diligence-issue-extraction with an optional argument of a VDR folder path or category name.
Be specific about the folder or category when you can. A request naming a folder gives the agent a clear starting point for the inventory step.
- "Review the data room for the Acme deal"
- "Extract issues from /02-Contracts"
- "Run a diligence review on the IP folder"
- "What's in the VDR?"
- "Here are the documents in /03-IP, find the issues"
/diligence-issue-extraction [VDR folder path or category name]Read the output
The skill produces a VDR inventory table with request category, VDR folder, document count and status, followed by a list of gaps. Findings follow the issue template in your CLAUDE.md. If your firm's seed memo uses that template, the skill uses it exactly. If the seed memo used bullets, the output uses bullets.
Findings are given a severity level under your house scheme. If the house scheme is red, yellow and green, red marks issues that affect deal value or structure.
- Inventory table: request category, VDR folder, docs, status
- Gaps: request categories with no VDR content, which need a follow-up request
- Each finding: issue title, category, severity, documents, finding, recommendation
- Citations keep their source tag so unverified ones can be checked first
Issue #N: [Title]
Category: [request list category]
Severity: [level per house scheme]
Documents: [VDR path + doc name]
Finding: [what the document says and why it matters]
Recommendation: [price adjustment / indemnity / consent required / rep & warranty / walk]Safety notes: scripts and network access
The skill does not run scripts. It does use the network. If a VDR MCP connector or a legal research connector is configured, the agent sends requests through it to retrieve the index, documents or citations. Check which connectors are attached before you start a review, and confirm that the data room host and research tool are ones your firm has approved for that deal.
The skill also tells the agent not to fill gaps silently. If a research query returns few or no results, the agent reports that and offers options, including web search, which is tagged [web search — verify]. Statute descriptions the agent cannot confirm from retrieved text should be marked [statute unretrieved — verify] rather than guessed.
- No scripts are run by this skill
- Network use happens through any VDR or legal research connectors you configure
- Web search is only used if you choose that option, and its results are tagged for verification
- Matter files are written under the corporate-legal matters folder; other matters are not read unless cross-matter context is on
Tips for a useful first-pass review
Set your materiality thresholds in CLAUDE.md and deal-context.md before the first run. The skill reviews documents top-down until it hits the threshold, so vague thresholds lead to vague coverage.
Review the gaps list first. A request category with no VDR content is often the most important follow-up for the deal team.
Check tagged citations, starting with those marked verify. The source states that citations carrying verify tags have higher fabrication risk.
- Define thresholds for contracts (for example by stated value) before running
- Start with the inventory and gaps, then review material categories
- Use folder names or category names in your request to narrow the scope
- Treat every finding as a draft for lawyer review
What people use it for
- Running a first-pass review of a large data room to find the documents that matter for a deal
- Building a consistent issue list from your firm's own categories and materiality settings
- Spotting change-of-control and assignment clauses in material contracts before signing
- Identifying IP ownership gaps, such as missing founder or employee assignments
- Producing a list of request categories with no VDR content for a follow-up request
Frequently asked questions
How do I install the diligence issue extraction skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for all projects, or into a project's .claude/skills/ folder for one project. You can also upload it in the Claude apps under Settings → Capabilities → Skills.
What do I need before the skill works?
The source requires the Claude for Legal corporate-legal configuration files, CLAUDE.md and deals/[code]/deal-context.md, under ~/.claude/plugins/config/claude-for-legal/corporate-legal/. The skill depends on them to run correctly. VDR and legal research connectors are optional.
Is the diligence issue extraction skill free?
The source does not state pricing or licence fees for the skill. It is listed under the Apache-2.0 licence. Check the licence file for its terms.
Can the skill give a final legal opinion?
No. The source describes the output as an AI draft that needs lawyer review. It also warns that unverified citations carry fabrication risk, so findings should not be used as final legal opinions or sign-off.
Does the skill run scripts or access the internet?
It runs no scripts. It does use the network through any VDR or legal research connectors you configure, and web search is used only if you choose that option.
What should I do if the skill returns few or no research results?
The skill reports the number of results and stops rather than filling the gap silently. You can broaden the query, try a different research tool, search the web with results tagged [web search — verify], or flag the item as unverified.
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 or law firm lawyers who want a consistent first-pass issue list across a large data room, provided they have set up the required configuration files. It is not a substitute for lawyer review, final opinions or sign-off. If you do not have the CLAUDE.md and deal-context.md files, or you need a finished legal opinion, choose a different approach.
Download the skill