How to use the escalation-flagger agent skill for contract approvals

What is an agent skill and what does this one do?

An agent skill is a folder containing a SKILL.md file that teaches an AI assistant how to handle a specific task. The assistant loads the skill when a request matches its description.

This skill reads the Escalation section of your practice profile at ~/.claude/plugins/config/claude-for-legal/commercial-legal/CLAUDE.md. It characterizes the contract issue as a dollar threshold, a term deviation, an automatic trigger or a business decision, then matches it to the matrix and names the approver. It then drafts a message containing the contract quote, the playbook position, options with a recommendation and a decision-by date.

• It does not send anything. It drafts and shows the message, and the lawyer reviews and sends it.
• If the matrix does not name an approver for a situation, the skill says so instead of inventing one.
• If the Escalation section is missing or vague, the skill reports that the practice profile needs editing.

Install the escalation-flagger skill

Download the ZIP for this skill from this page. It contains the SKILL.md file.

Option A: unzip it into your personal skills folder, ~/.claude/skills/. Option B: unzip it into a project's .claude/skills/ folder so it applies only to that project.

Option C: in the Claude apps, open Settings, go to Capabilities, then Skills, and upload the ZIP.

• Confirm the unzipped folder contains SKILL.md at its top level.
• Restart or start a new session if the skill does not appear in your agent.

Set up your escalation matrix before you run it

The skill depends on a configured practice profile. Without it, the skill has nothing to match against and will report the files as missing.

The Escalation section should contain a table showing who can approve what, the dollar or risk threshold, who it escalates to and the channel used. Automatic escalation triggers, such as unlimited liability or IP assignment, should also be listed. The playbook sections it reads are the Sales-side playbook and Purchasing-side playbook.

• Required file: ~/.claude/plugins/config/claude-for-legal/commercial-legal/CLAUDE.md
• Required content: an Escalation section and playbook sections with stated positions, fallbacks and automatic escalation items.
• Example matrix rows in the source: Paralegal for standard terms under $50K, Counsel for non-standard terms within fallbacks under $500K, and GC for everything else.

Trigger the skill with natural requests

You can invoke the skill with the slash command or simply describe what you need. The skill's description lists phrases that should activate it.

Phrases from the source that trigger it include "who needs to approve this", "escalate this", "does this need GC sign-off" and "route this for approval". It also activates when another skill finds an issue that exceeds the reviewer's authority.

• Use the slash command form with a description: /commercial-legal:escalation-flagger followed by the issue.
• Or reference a review memo: give the memo file name and the issue, such as a specific section of the indemnity clause.

Read the output and decide what to send

The draft follows a fixed template: escalating to, via, urgency, the issue in plain English, the exact contract quote, the playbook quote, three options (accept, push back with counter-language, or walk) and a recommendation. Each option includes a one-line rationale.

When the skill is uncertain whether a term sits inside the fallbacks, it still escalates and flags the specific question for the approver. You review the draft, edit it and send it yourself.

• Check the recommendation against your own judgment before sending.
• If the team uses a ticket system or CLM approval workflow, log the escalation there. Otherwise, note in the review memo who received it and when.

Safety notes: scripts and network

According to the skill's listing, it runs no scripts and uses no network access. It reads the local CLAUDE.md file and the review material you provide, and it writes a draft for you to read.

It is not a source of final legal judgment. The drafted recommendation is a starting point and must be checked by a qualified lawyer before anything is sent.

• If the matter workspace is enabled, the skill writes outputs to the matter folder under ~/.claude/plugins/config/claude-for-legal/commercial-legal/matters/.
• It does not read other matters' files unless cross-matter context is turned on.

Tips for better results

The quality of the output depends on the quality of your matrix and playbook. Vague thresholds produce vague routing. Give the skill the exact clause, the counterparty name and the agreement type when you can.

• Name the counterparty and agreement type in your request.
• Quote the clause or give its section number, for example §8.2 indemnity carveouts.
• Update the playbook when approvers keep receiving escalations that should have been handled at the reviewer level.
• Expect extra escalations when a term is ambiguous. The skill is designed to escalate when in doubt.