How to use the Deposition Prep Outline agent skill

What the deposition-prep skill makes the agent do

When you ask for deposition prep for a witness, the agent loads your firm's case theory and key facts from a local plugin config file. It then pulls documents authored by or mentioning the witness from your eDiscovery platform and builds an outline with four parts: background, key documents, topics tied to the case theory, and impeachment material.

The skill also follows a structure. It asks about the witness's role and why you are deposing them, and it checks the witness's posture (adverse, friendly, neutral third-party, or corporate representative) before drafting questions, because the question style differs for each.

• Background on the witness and their relationship to the case
• Key documents tied to the witness
• Topics organized around the case theory
• Impeachment material drawn from the record

Install the deposition-prep skill

Download the ZIP from the anthropics/claude-for-legal page for litigation-legal/skills/deposition-prep. The skill ships as a single file, SKILL.md.

For a personal install, unzip the folder into ~/.claude/skills/. For a single project, unzip it into the project's .claude/skills/ folder.

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

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

Set up the config files before you run it

The skill reads two files from the litigation-legal plugin configuration. The first holds your case theory and key facts. The second is the matter log, which the matter-intake step creates. The skill will not build an outline for a matter that is not in the log.

The source does not name a specific eDiscovery product, so you need access to one that your team already uses.

• ~/.claude/plugins/config/claude-for-legal/litigation-legal/CLAUDE.md holds the case theory (theory, pivot fact, key facts for and against)
• ~/.claude/plugins/config/claude-for-legal/litigation-legal/matters/_log.yaml is the matter log, created by /litigation-legal:matter-intake
• Access to an eDiscovery platform for pulling the witness's documents

Trigger the skill with natural requests

The skill activates on requests such as the ones below. You can also use the slash form /deposition-prep with a witness name as the argument.

Name the witness clearly. The agent needs to know who the outline is for and what you need from them, since that shapes the centerpiece of the outline.

• "depo prep for [witness]"
• "build a depo outline"
• "prepare for [name]'s deposition"
• /deposition-prep [witness name]

Work through the outline with the agent

Start by answering the agent's questions about the witness's role, their relationship to the case, and the purpose of the deposition. Confirm the witness's posture, since a hostile witness and your own client need different question styles.

Review the draft against the record. Quotes must be verbatim or marked for verification, and pinpoint citations must support the full proposition they are attached to. Where the agent marks a placeholder such as [verify exact quote], check it against the transcript before you rely on it.

The skill advises keeping to three or four topics that matter, leading with the strongest confrontation, and flagging where to cut if the record is deep.

• Confirm witness posture before drafting questions
• Check every [verify] placeholder against the transcript
• Trim to the three or four topics that matter

Know the guardrails and safety limits

The skill enforces several rules. It refuses to build an outline for a matter that has not been run through intake, because intake runs the conflicts check. Before producing output, it checks whether the destination is inside the privilege circle and asks before sending output to a public channel, a counterparty, a vendor, or a client. It will not draft a witness statement in the witness's voice.

For England & Wales, the skill notes that PD 57AC governs trial witness statements. It will prepare question prompts, capture the witness's own words, generate the list of documents shown, run a compliance checklist on a statement the witness drafted, and draft the solicitor's certificate. It will not write the evidence.

Safety notes from the listing: the skill does not run scripts, but it does use the network, since it pulls documents from an eDiscovery platform and may query a legal research tool. Its output is a first draft that needs review by a qualified lawyer.

• Refuses unintaken matters (conflicts gate)
• Asks before sending to non-privileged destinations
• Will not write a witness statement in the witness's voice
• Runs no scripts; uses the network for eDiscovery and research queries
• Outputs need review by a qualified lawyer and citations must be checked against the transcript