Skip to content

Guide · Agent skill

How to use the Internal Investigation Framework agent skill

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

The internal-investigation skill is a reference framework for an AI assistant helping a legal team run a workplace or corporate internal investigation, from intake through a final memo. It is written for counsel and is loaded by other commands rather than invoked on its own. After reading this guide you will know what the skill does, how to install it, how to trigger it with natural requests, and what safety points to check before using it on a real matter.

Internal Investigation Frameworkanthropics/claude-for-legal · Apache-2.0

Quick start

  1. 1Download and unzip. Download the ZIP from this page and unzip it to get the internal-investigation folder.
  2. 2Place the skill. Move the folder to ~/.claude/skills/ for personal use, or to a project's .claude/skills/ folder.
  3. 3Confirm the plugin and config. Make sure the claude-for-legal employment-legal plugin is installed and the practice-level CLAUDE.md is in ~/.claude/plugins/config/claude-for-legal/employment-legal/.
  4. 4Confirm attorney direction. Before opening a matter, confirm the investigation is attorney-directed, since privilege depends on it.
  5. 5Open a matter. Ask the assistant to open an investigation and answer the intake questions in one block.
  6. 6Answer the representation flags. Disclose any union representation or public-employer status so the Weingarten or Garrity research is triggered before interviews.
  7. 7Review the checklist and query the log. Check the generated sources checklist, mark items N/A where needed, and ask questions against the log as the matter progresses.

What the internal-investigation skill makes the agent do

The skill gives the assistant a structured workflow for managing an internal investigation. It checks whether the matter is attorney-directed, opens a structured investigation log, builds a checklist of sources to gather, and tracks what has and has not been gathered.

It also helps process document collections to surface what matters, answer questions against the log, draft a memo, and produce audience summaries. Before interviews begin, it flags legal issues such as union representation (Weingarten) and public-employer compelled statements (Garrity).

  • Intake: asks about the allegation, complainant, respondent, timeframe and investigation type
  • Investigation log: a structured record of entries, issues and evidentiary gaps
  • Sources checklist: generated from the investigation type (HR, financial, executive, whistleblower, or other)
  • Document processing, Q&A against the log, memo drafting and audience summaries

Install the internal-investigation skill

Download the ZIP file from this 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.

You can also upload the skill in the Claude apps under Settings → Capabilities → Skills.

The skill is designed to run inside the claude-for-legal employment-legal plugin, so the plugin must be installed for its commands to load it.

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

Set up the practice-level configuration

The skill reads practice-level settings from ~/.claude/plugins/config/claude-for-legal/employment-legal/. The skill expects a practice-level CLAUDE.md configuration in that location, including an escalation table and an output header that reflects your role.

If you use matter workspaces, the CLAUDE.md includes a Matter workspaces section. If it is disabled (the default for in-house users), the matter machinery is skipped and the skill uses practice-level context.

~/.claude/plugins/config/claude-for-legal/employment-legal/CLAUDE.md

Trigger the skill with natural requests

The skill is loaded by these commands: /employment-legal:investigation-open, /employment-legal:investigation-add, /employment-legal:investigation-query, /employment-legal:investigation-memo and /employment-legal:investigation-summary. It is not invoked directly, but natural phrasing also works for opening a matter, such as "open an investigation" or "start an investigation into".

When you open a matter, the assistant asks for the allegation, complainant, respondent, timeframe and whether the matter is attorney-directed, all in one block.

  • "Open an investigation into alleged expense fraud by a regional manager."
  • "Start an investigation into the complaint from the warehouse team about retaliation."
  • /employment-legal:investigation-open to begin intake for a new matter

Work through the intake and flags

Intake asks whether any complainant, respondent or witness is represented by a union or covered by a collective bargaining agreement. If so, the skill flags Weingarten research and says not to interview until the protocol is adjusted.

It also asks whether the company is a public employer or acts under color of state law. If so, it flags Garrity research, since compelled statements in public-sector investigations carry special use-immunity consequences.

  • Weingarten flag: union representation or collective bargaining coverage
  • Garrity flag: public employer or acting under color of state law
  • Do not begin interviews until the flagged protocol is adjusted

Safety notes: privilege, scripts and network

Marking does not create privilege. The skill's work-product header reflects intended protection, but whether an output is privileged depends on whether the investigation is attorney-directed, why documents were created, and how they are used or disclosed. Confirm attorney direction before opening a matter.

The skill runs no scripts and uses no network access, according to its listing. Its files are written to the matter folder under ~/.claude/plugins/config/claude-for-legal/employment-legal/, so store them where privileged materials live and limit distribution deliberately.

  • If the investigation is not attorney-directed, the skill's default labelling may be misleading
  • Forwarding files to non-attorneys, or cc'ing HR without scoping, can waive privilege
  • Privilege is a legal judgement; the tool does not settle it

Tips for getting useful output

Give the intake block complete answers, especially the investigation type and the representation questions, because they determine the sources checklist and which flags fire.

Review the generated sources checklist and mark items not applicable rather than deleting them, so the gap tracking stays accurate.

  • Name the investigation type precisely so the checklist fits the matter
  • Ask the assistant to query the log rather than re-reading raw documents
  • Keep the header on every file the skill produces

What people use it for

  • Opening a structured log for a harassment or retaliation complaint in an HR investigation
  • Generating a sources checklist for an expense-fraud or procurement investigation
  • Flagging Weingarten representation rights before interviews with a unionised team
  • Flagging Garrity issues before interviews in a public university investigation
  • Drafting a privileged memo and audience summaries from an existing investigation log

Frequently asked questions

Is the internal-investigation skill free to use?

The source does not state pricing or licence terms beyond the licence listed on this page, which is Apache-2.0. Check the licence file in the repository for the terms that apply.

How do I install the internal-investigation skill?

Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use or a project's .claude/skills/ folder. You can also upload it in the Claude apps under Settings → Capabilities → Skills.

What does the internal-investigation skill require?

It requires Claude Code or a compatible agent with the claude-for-legal employment-legal plugin installed, plus a practice-level CLAUDE.md at ~/.claude/plugins/config/claude-for-legal/employment-legal/. It runs no scripts and uses no network.

Can I use this skill for my own workplace complaint?

Not on its own. The skill is written for counsel and is not invoked directly, so it will not answer a personal legal question. Anyone conducting an investigation without attorney direction should not rely on its privilege labelling.

Does the skill make my investigation files privileged?

No. The skill states that marking does not create privilege. Whether outputs are privileged depends on whether the investigation is attorney-directed, the purpose of the documents, and how they are used or disclosed.

What should I do if a witness is in a union?

Tell the assistant during intake. The skill flags Weingarten research and says not to conduct interviews until the protocol is adjusted, so counsel should review the applicable rules first.

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 outside employment counsel who want a structured, labelled workflow for investigations, and legal ops or HR-legal teams standardising how files are organised. It is not suited to anyone running an investigation without attorney direction, or to non-lawyers seeking advice on their own complaint; in those cases, pick a general legal resource or consult a lawyer directly.

Download the skill