due-diligence
Structures and runs a diligence process with a workstream request list, a findings tracker, and a red-flag log.
SKILL.mdpacks/investment-banking/skills/due-diligence/SKILL.md
GitHubDue Diligence Agent
When to use
Use this when you must structure and run a diligence process on a target, whether buy-side confirmatory diligence or sell-side vendor diligence. It is the right tool the moment a deal moves past a signed LOI or a data room opens. Reach for it when you need to turn a vague "do diligence" instruction into an organized, owner-assigned process.
What it does
It produces a diligence request list organized by workstream and a findings tracker that logs every issue with severity, owner, and status, plus a running red-flag log that surfaces deal-threatening items early.
Method
Diligence checklist and findings log, worked step by step.
Frame the scope.
- Confirm deal type, sector, materiality thresholds, and the investment thesis to prove or disprove.
- Pitfall: an unbounded request list stalls the process; anchor every ask to a thesis question.
Build the request list by workstream: commercial, financial, legal, tax, operational, HR, and IT.
- Under each, list specific documents and data to pull from the data room or management.
- Commercial: customer concentration, churn, pipeline. Operational: capacity, supply chain, key facilities.
Prioritize quality of earnings in the financial workstream.
- Revenue recognition, one-time and non-recurring items, working capital trends, and normalized EBITDA with a clean add-back bridge.
- Watch for aggressive capitalization or channel stuffing that flatters earnings.
Cover legal, tax, HR, and IT.
- Legal and tax: material contracts, change-of-control clauses, litigation, IP ownership, open tax positions.
- HR: key-person dependencies and retention. IT: systems, security posture, and integration debt.
Assign each request an owner and a due date.
- Track receipt status: outstanding, received, under review, or cleared.
- Good practice: a single owner per item, never a team, so accountability is clear.
Log findings as documents arrive.
- Capture description, workstream, severity (high, medium, low), potential deal impact (price, structure, walk-away), owner, and status.
- Severity is about deal impact, not effort: a small dollar item with a legal walk-away trigger is high.
Escalate high-severity findings to the red-flag log.
- Add a one-line thesis on why it threatens value and a recommended mitigation (price adjustment, escrow, indemnity, or condition to close).
Roll findings into a diligence summary.
- Tie each material issue to a proposed action in the purchase agreement or valuation, so diligence directly informs deal terms.
Inputs
- Deal type (buy-side or vendor) and the target sector
- The core investment thesis to prove or disprove
- Data room access or a document index
- Materiality thresholds and the diligence timeline
- Names of workstream owners on the deal team
- Any prior findings or known issues to carry forward
- Purchase agreement stage, so findings can be routed to terms
Output format
- A request list grouped by the seven workstreams, each item with an owner and status
- A findings tracker as a prose list, each finding naming its workstream, severity, deal impact, owner, and status
- A red-flag log: a short ranked list of the highest-severity items, each with a recommended mitigation
- A diligence summary mapping material issues to purchase-agreement or valuation actions
- All of the above described in prose, never rendered as literal markdown tables
Example
For Meridian Foods, a buy-side process, the financial workstream flags a high-severity finding: 38 percent of revenue comes from a single grocery chain whose contract carries a change-of-control termination right. Deal impact is rated high and routed to the red-flag log, with a recommended mitigation of a specific indemnity plus a portion of consideration held in escrow until the contract is reaffirmed. A medium finding notes 4 million of owner add-backs in normalized EBITDA that thin the defensible earnings base. A low finding flags an expiring facility lease, tracked but not deal-threatening, and the summary ties the concentration issue directly to a proposed purchase-price adjustment.