Credit Agreement Analysis with AI: Covenants, Baskets, and Flex Provisions

Why Credit Agreement Review Is the Bottleneck in Direct Lending

A middle-market credit agreement for a direct lending deal runs 150 to 300 pages before schedules and exhibits. By the time you include the defined terms, the representations and warranties, the affirmative and negative covenant packages, the events of default, the waterfall mechanics, and the amendment provisions, you have a document that takes an experienced credit attorney two to three days to review thoroughly and a credit analyst another day to translate into actionable monitoring terms. For a fund closing 15 to 20 deals per year, the document review burden is substantial — and errors in reading covenant definitions or missing a basket carve-out can have real consequences when a borrower approaches the boundaries of its permitted activities.

Claude accelerates credit agreement review at three distinct points: initial covenant extraction and definition parsing, ongoing basket headroom quantification, and amendment analysis when a borrower requests a change. The tool does not provide legal opinions — that remains counsel's work — but it dramatically reduces the time to extract, organize, and test the economic terms that drive lender decision-making. This guide covers how direct lending analysts and portfolio managers can use Claude across the full credit agreement lifecycle, from initial deal review through portfolio monitoring.

For context on the broader workflow, see our guides on covenant compliance monitoring and private credit portfolio intelligence.

What Credit Agreement Review Actually Involves

Credit agreements are structured around five analytical layers, and the review focus depends entirely on who is doing the reviewing. An arranger reviewing the deal for syndication focuses on structure, pricing flex, and whether the documentation matches the commitment letter terms. An existing portfolio lender doing quarterly monitoring focuses on covenant headroom, basket utilization, and whether the borrower's recent actions have consumed capacity that might be needed later. A prospective secondary buyer buying a loan in the market focuses on the transfer restrictions, pro rata requirements, and whether the existing covenant package is at market or tight relative to comparable credits.

The five layers that matter most for credit analysis are: (1) financial maintenance covenants — the leverage, coverage, and liquidity tests tested quarterly; (2) negative covenants — the restrictions on what the borrower can do, with the permitted exceptions (baskets); (3) representations and warranties — the factual statements the borrower makes at closing and reaffirms on each draw; (4) events of default — the specific triggers that give lenders acceleration rights; and (5) amendment and consent mechanics — who has to agree to change the agreement and what can and cannot be changed by a simple majority.

The most analytically intensive part of the review is the negative covenant basket analysis. A typical middle-market credit agreement has 15 to 25 separate baskets across the restricted payments, permitted debt, permitted investments, permitted liens, and asset sale covenant categories. Each basket has a dollar threshold (fixed or grower), sometimes an incurrence test condition, and often a builder basket that grows over time. Understanding the current state of basket utilization — what the borrower has consumed and what remains — is essential for assessing the borrower's future strategic optionality and, from the lender's perspective, the risk of covenant arbitrage.

Financial Covenant Extraction and Testing

The Consolidated EBITDA definition is the most important defined term in any leveraged credit agreement, and it is almost never straightforward. The base calculation — net income plus interest, taxes, depreciation, and amortization — is standard. Everything after that is negotiated: what non-cash charges are addable, whether management fees paid to the sponsor count (almost always yes, with a cap), whether restructuring charges are permitted and capped, whether pro forma cost savings from acquisitions are addable and for how long, and whether there is a general "non-recurring" bucket that the borrower can use as a catch-all. The LTM measurement period mechanic also matters: rolling four fiscal quarters is standard, but some agreements use a run-rate from the most recent quarter multiplied by four, which can produce very different results for a growing or declining business.

The leverage ratio test flows directly from the EBITDA definition. A covenant stated as "Maximum Consolidated Total Net Leverage Ratio of 4.50 to 1.00" requires computing Consolidated Total Net Debt (usually all debt minus unrestricted cash, sometimes with a cap on nettable cash) divided by LTM Consolidated EBITDA as defined. Getting the denominator wrong by $1M to $2M in a $20M EBITDA company changes the leverage ratio by 0.05 to 0.10 turns — meaningful when the covenant is set at 25 to 30 basis points of cushion to closing leverage.

• "I am reviewing the credit agreement for Cascade Distribution Group, a specialty industrial distributor. Below is the complete Consolidated EBITDA definition from Section 1.01 of the credit agreement. Parse this definition and produce a structured summary: (1) list every permitted add-back category, the section reference, and whether there is a dollar cap; (2) identify any non-recurring or catch-all add-back categories and the cap amount; (3) explain the Pro Forma Cost Savings mechanic — what conditions must be met before synergies can be included, and for how many months post-acquisition; (4) identify any items that are added back only if non-cash, with an explanation of the cash-reversal mechanic; (5) explain the LTM measurement period as defined — rolling four quarters, partial period annualization, or other. I will use this as my definitive EBITDA calculation reference for quarterly monitoring. [CREDIT AGREEMENT EBITDA DEFINITION: 'Consolidated EBITDA means, with respect to the Borrower and its Restricted Subsidiaries for any period, Consolidated Net Income for such period plus (to the extent deducted in calculating Consolidated Net Income and without duplication): (i) Consolidated Interest Expense; (ii) provision for income taxes; (iii) depreciation and amortization; (iv) any non-cash charges, including non-cash stock-based compensation, provided that if any such non-cash charge represents an accrual for potential future cash payments, such future cash payment shall be subtracted from Consolidated EBITDA in the period paid; (v) management, monitoring, and consulting fees paid to the Sponsor or its Affiliates not to exceed $2,500,000 per fiscal year; (vi) non-recurring, extraordinary, or one-time charges, costs, and expenses not to exceed the greater of $6,000,000 and 10% of Consolidated EBITDA in any trailing twelve-month period; (vii) Pro Forma Cost Savings expected to be realized within 18 months of a Permitted Acquisition, as certified by a Responsible Officer in a certificate delivered to the Administrative Agent; minus (viii) non-cash income items increasing Consolidated Net Income.']"
• "Quarterly covenant compliance test for Cascade Distribution Group, Q2 2026. Credit agreement covenants: (A) Maximum Total Net Leverage Ratio: 4.25x (steps to 4.00x at Q4 2026, 3.75x at Q2 2027); (B) Minimum Interest Coverage Ratio: 2.50x; (C) Minimum Liquidity (unrestricted cash plus undrawn revolver availability): $7,500,000. LTM financials for the four quarters ending Q2 2026 (Q3 2025 through Q2 2026): Q3 2025: Revenue $38.2M, Net Income $2.1M, Interest Expense $3.2M, Taxes $0.7M, D&A $1.4M, Sponsor management fee $0.6M, one-time consulting $0.8M; Q4 2025: Revenue $41.5M, Net Income $2.9M, Interest Expense $3.2M, Taxes $1.0M, D&A $1.4M, Sponsor management fee $0.6M; Q1 2026: Revenue $36.4M, Net Income $1.6M, Interest Expense $3.3M, Taxes $0.5M, D&A $1.5M, one-time warehouse closure costs $1.4M (cash); Q2 2026: Revenue $39.8M, Net Income $2.4M, Interest Expense $3.3M, Taxes $0.8M, D&A $1.5M. Balance sheet at Q2 2026: Term loan outstanding $68.5M, revolver $15M ($3M drawn), unrestricted cash $4.2M. Calculate: (1) LTM Consolidated EBITDA under the credit agreement definition (state which add-backs are permitted and which are not, with reasoning); (2) Consolidated Total Net Debt; (3) each covenant ratio vs. threshold; (4) headroom on each covenant as a ratio and a percentage; (5) produce a compliance certificate table I can present to the borrower for sign-off."

Negative Covenant Analysis: Baskets, Grower Baskets, and Builder Baskets

The negative covenant section of a credit agreement restricts four main categories of borrower behavior: incurring additional debt, making restricted payments (dividends, stock buybacks, subordinated debt payments), making investments, and selling assets. Each restriction comes with a set of exceptions — the "baskets" — that define the space within which the borrower can operate without lender consent. Understanding which baskets are available, how much capacity remains in each basket, and whether basket usage has been properly tracked is the core of negative covenant portfolio analysis.

Fixed baskets are straightforward: a $20M permitted investment basket means the borrower can make up to $20M of investments (broadly defined) without triggering the restricted investments covenant. Grower baskets are more complex: a basket defined as "the greater of $15,000,000 and 15% of Consolidated EBITDA" means the basket grows as EBITDA grows. At $20M EBITDA, the basket is $15M (the fixed floor). At $30M EBITDA, it grows to $4.5M greater than the floor, giving $15M still. At $120M EBITDA, the 15% component dominates at $18M. Grower baskets effectively give growing companies more room as they scale.

Builder baskets — sometimes called the "cumulative credit" — are a separate mechanic. They accumulate over time based on a percentage of excess cash flow (typically 50% of Consolidated Net Income) generated since closing, and are available for restricted payments, investments, or debt prepayments. The builder basket balance must be tracked across every use since closing, which is one of the most commonly mis-tracked items in portfolio monitoring. A borrower that has built up a $12M builder basket and quietly used $8M of it for a management equity purchase has $4M remaining — not $12M.

• "I need to quantify the current restricted payment basket headroom for Cascade Distribution Group. The credit agreement restricted payments covenant reads as follows: 'No Restricted Payment shall be made except: (i) the General RP Basket: Restricted Payments in an aggregate amount not to exceed the greater of (A) $12,000,000 and (B) 12.5% of Consolidated EBITDA for the most recently ended Test Period (the 'General RP Grower Basket'), provided no Default or Event of Default exists; (ii) the Builder Basket: Restricted Payments in an aggregate amount not to exceed the Cumulative Credit, defined as 50% of Consolidated Net Income for the period from the Closing Date through the most recently ended fiscal quarter, less the aggregate amount of all prior Restricted Payments made under this clause (ii) and all Permitted Investments made under the Investment Builder Basket; (iii) the Sponsor Fee Basket: management fees paid to the Sponsor not to exceed $2,500,000 per year.' Given: LTM Consolidated EBITDA $19.4M; Consolidated Net Income since closing (Q3 2022 through Q2 2026, 16 quarters) cumulative total $18.6M; prior restricted payments under the Builder Basket: $4.2M management equity purchase (Q1 2024), $1.8M sponsor dividend (Q4 2025); no General RP Basket usage since closing. Calculate: (1) current General RP Grower Basket capacity; (2) current Builder Basket balance; (3) total restricted payment capacity across all baskets; (4) confirm whether the sponsor management fee basket is tracked separately. Flag whether prior basket usage has been properly documented in the compliance record."
• "Permitted debt basket analysis for Cascade Distribution Group. I need to quantify all available debt capacity outside the existing term loan and revolver. The credit agreement permitted debt baskets are as follows: '(a) Existing ABL Facility up to $15,000,000; (b) Capital Lease Obligations not to exceed the greater of $8,000,000 and 8% of Consolidated EBITDA; (c) Purchase Money Indebtedness not to exceed the greater of $5,000,000 and 5% of Consolidated EBITDA; (d) General Debt Basket not to exceed the greater of $10,000,000 and 10% of Consolidated EBITDA; (e) Ratio Debt permitted if, after giving Pro Forma Effect thereto, the Total Net Leverage Ratio does not exceed 3.75 to 1.00 (the Ratio Debt Incurrence Test).' Current utilization: ABL $3M drawn; capital leases $2.4M; purchase money debt $0. Current LTM Consolidated EBITDA $19.4M. Current Total Net Leverage Ratio 3.52x. Calculate: (1) remaining capacity under each fixed/grower basket; (2) whether Ratio Debt is currently available given the 3.52x leverage and the 3.75x incurrence test; (3) maximum total additional debt capacity combining all available baskets; (4) what leverage ratio would result if the borrower fully utilized all available basket capacity. This analysis supports a request from the borrower to finance a $12M equipment purchase — advise on which basket(s) they would use."

Events of Default Review

The events of default section defines the conditions under which lenders gain acceleration rights — the ability to declare all debt immediately due and payable. For credit analysts, the key events of default to understand are: (1) payment default on principal or interest, including the applicable grace period (usually 3 to 5 business days for interest, immediate for principal); (2) covenant default, which typically carries a 30-day cure period for financial covenants and longer cure periods (60 to 90 days) for reporting covenants; (3) cross-default and cross-acceleration, which trigger default if the borrower defaults on other debt above the cross-default threshold; (4) material adverse change, which is subjective and rarely used in practice but must be understood; and (5) change of control, which typically defines control as ownership above 50% shifting to a non-permitted holder.

The cross-default clause deserves particular attention in a portfolio context. A borrower that has a $50M cross-default threshold and is approaching default on a $55M subordinated note is on the verge of triggering default across the entire debt stack, not just the subordinated instrument. Lenders need to track the borrower's entire debt profile — including equipment financing, capital leases, and any subordinated instruments — to understand the cross-default exposure map.

• "Review the Events of Default section of this credit agreement and produce a structured analysis: (1) list every event of default, its section reference, and the applicable notice period and cure period; (2) identify the cross-default threshold amount and whether it applies to payment defaults only or also to acceleration events (cross-acceleration vs. cross-default distinction); (3) describe the Material Adverse Change definition and note any carve-outs from the MAC definition (changes in general economic conditions, industry conditions, changes in law — are these carved out from the MAC definition?); (4) explain the change of control definition and whether a PE-to-PE sale is permitted without lender consent; (5) identify which events of default have cure periods and which are immediate (no cure). Flag the three events of default that pose the most practical risk given the borrower's current financial profile. [Insert credit agreement Events of Default section]"

Amendment and Waiver Analysis

Credit agreement amendments and waivers are a routine part of portfolio management for any direct lending fund. Understanding the amendment mechanics — who must consent and for what — is critical both when the borrower requests a change and when the fund is considering acquiring a secondary position and needs to know what changes have been consented to since closing.

The Required Lender definition is the starting point. Most credit agreements define Required Lenders as holders of more than 50.1% of the aggregate outstanding loans and unused commitments. Financial covenant amendments almost always require Required Lender consent. However, certain provisions — sometimes called "sacred rights" or "all-lender provisions" — require unanimous consent: reducing the applicable interest rate, forgiving principal, extending the maturity date, releasing the collateral agent's lien on all or substantially all collateral, and releasing a guarantor. In a club deal with five lenders where one holds 40% of the commitment, an amendment requiring super-majority (66.7%) consent gives that large holder significant blocking power — and significant leverage in the amendment negotiation.

• "Cascade Distribution Group has requested a covenant amendment: (1) deferral of the Q4 2026 leverage step-down from 4.25x to 4.00x by two quarters to Q2 2027, citing a $4M EBITDA shortfall from a delayed customer contract; and (2) a permanent increase in the non-recurring add-back cap from $6,000,000 to $8,500,000. The credit agreement has 4 lenders: Lender A holds 40%, Lender B holds 30%, Lender C holds 20%, Lender D holds 10%. The Required Lender threshold is 50.1%. Analyze: (1) what consent is required for each requested amendment — is this a Required Lender vote or does it require unanimous consent? (2) which combination of lenders must vote yes to reach Required Lender threshold, and which combinations can block? (3) what amendment fee and tightened terms is market practice for a leverage step-down deferral of this type (cite typical range: 12.5bps to 25bps of outstanding principal, plus possible pricing flex of 25-50bps)? (4) draft the key terms of a proposed amendment and fee letter to present to the borrower in response to their request."

Flex Provision Analysis

In broadly syndicated deals and some larger club deals, the commitment letter includes flex rights allowing the arranger to modify pricing, structure, or terms if the syndication requires it. Understanding the flex provisions is important for any lender reviewing a deal at the commitment letter stage — before the final credit agreement is signed — because flex can materially change the economics and structure of the deal.

Flex rights typically include: pricing flex (the arranger can increase the spread by up to 50-75bps above the commitment letter rate if investors demand wider pricing), structure flex (OID can be increased by up to 1-2 points), and in extreme cases, structural flex (converting term loan proceeds to revolving credit, adding a second lien tranche, or changing amortization). Understanding the total amount of flex remaining when you are reviewing a commitment letter — i.e., how much worse the final deal terms could get before you breach your investment mandate — is an important part of commitment-stage diligence.

• "I am reviewing the flex provisions in the commitment letter for a $240M unitranche facility arranged for Meridian Industrial Group. The commitment letter states: 'Pricing Flex: The Arrangers may increase the Applicable Margin by up to 75 basis points above the Initial Margin of SOFR+575. OID Flex: The Arrangers may increase the OID from 2.0% to a maximum of 3.5%. Structure Flex: If the facility is not fully subscribed at the Initial Terms, the Arrangers may add a second lien tranche of up to $30M to replace a corresponding portion of the unitranche, with pricing at Initial Margin plus 200bps.' Analyze: (1) what is the maximum all-in yield to a lender if full pricing and OID flex is exercised, assuming 5-year term for OID amortization? (2) what is the worst-case total cost of capital to the borrower under full flex? (3) if structural flex is exercised and $30M shifts to second lien, what does the resulting capital structure look like and how does it change the effective blended cost? (4) as a direct lender being asked to commit $40M to the unitranche, what is the maximum all-in yield I can expect and what is the minimum (at initial terms)? Summarize in a flex sensitivity table."

Building a Credit Agreement Review Workflow with Claude

The most efficient setup for ongoing credit agreement work is a Claude Project dedicated to each portfolio company. At closing, paste the entire credit agreement into the project as reference documents (Claude can handle 200,000 tokens of context, enough for most middle-market credit agreements plus schedules). Configure the project instructions to tell Claude: "You are a credit analyst reviewing this credit agreement. Refer to the attached document for all definitions and covenant terms. Do not infer terms that are not in the document — if a definition is missing or ambiguous, flag it."

The quarterly workflow then runs in three steps. First, covenant extraction at closing (one-time): ask Claude to produce a master covenant summary with every financial covenant, every basket, and every defined term that feeds into the calculations. Second, quarterly testing: paste the quarterly financials and run the covenant calculation prompts. Third, as-needed amendment analysis: when the borrower requests a change, paste the amendment request and ask Claude to identify the consent mechanics, comparable precedent, and the appropriate counter-proposal.

For portfolio-scale covenant monitoring, see our guide on covenant compliance monitoring for private credit and the private credit portfolio intelligence framework. For the upstream credit underwriting workflow that precedes the credit agreement, see our guide on direct lending credit analysis.

Where to Start

Start with the financial covenant extraction prompt on your next deal review. Take 10 pages of the credit agreement — specifically the Consolidated EBITDA definition and the leverage and coverage covenant sections — paste them into Claude, and ask it to produce the master covenant calculation reference. Compare the output to what your deal team produced manually. In most cases, the Claude output will catch add-back nuances or cap mechanics that are easy to miss in a manual read. Once you have a validated extraction process, expand to the full negative covenant basket analysis. The basket headroom analysis is typically where the greatest time savings occur — a complete basket inventory that would take three to four hours manually takes 20 minutes with Claude. The Commercial Banking and Private Equity skill categories include credit agreement review templates for direct lending workflows.

Frequently Asked Questions

What is the difference between a maintenance covenant and an incurrence covenant?

A maintenance covenant must be satisfied continuously — tested each quarter regardless of whether the borrower is taking any action. A leverage ratio tested at 4.50x every quarter is a maintenance covenant. An incurrence covenant is only triggered when the borrower wants to do something specific: incur additional debt, make an acquisition, or pay a dividend. Many broadly syndicated loans have shifted to incurrence-only covenant packages (covenant-lite), while direct lending credit agreements for middle-market companies still commonly include financial maintenance covenants that provide quarterly monitoring triggers.

What is a grower basket in a credit agreement?

A grower basket is a negative covenant exception whose dollar capacity scales with the borrower's size — typically as "the greater of [$X] and [Y]% of Consolidated EBITDA." As EBITDA grows, the basket grows with it, giving larger companies more room to make investments, incur debt, or make restricted payments. Most modern credit agreements include both fixed baskets (flat dollar amount) and grower baskets as alternatives, with the borrower able to use whichever is larger. For credit analysts, grower baskets must be recalculated each quarter as EBITDA changes.

How does a cross-default clause work and why does the threshold matter?

A cross-default clause triggers an event of default under the current credit agreement if the borrower defaults on other debt exceeding a specified dollar threshold. A $50M cross-default threshold means only a default on debt above $50M triggers a cross-default — a missed lease payment on a $2M equipment loan does not. The threshold size matters significantly for lenders: a high threshold ($50M in a $70M term loan deal) provides little cross-default protection, while a low threshold ($5M) creates sensitivity to routine operational defaults on small obligations. Also watch for whether it is a cross-default (any default on other debt) vs. cross-acceleration (requires the other debt to be accelerated), since the latter is more borrower-friendly.

What provisions in a credit agreement require unanimous lender consent to amend?

Most credit agreements define certain "sacred rights" or "all-lender provisions" requiring unanimous (100%) consent: reducing the applicable interest rate or margins, forgiving or canceling principal, extending the stated maturity date, releasing all or substantially all collateral, releasing a material guarantor, and changing the pro rata payment mechanics. Financial covenant amendments — adjusting a leverage ratio threshold or adding an add-back — typically require only Required Lender (50.1%) consent. Structural changes like adding a new tranche or modifying waterfall priority generally require super-majority (66.7%) or unanimous consent depending on how the agreement is drafted.