dissenting,
I would dismiss for lack of subject matter jurisdiction. In deciding that the amount-in-controversy threshold is satisfied, the majority adopts the reasoning in Meridian Security Insurance Company v. Sadowski, 441 F.3d 536, 539 (7th Cir. 2006), to conclude that the rule against aggregation does not apply to this declaratory judgment action because “at the time of filing of the declaratory judgment complaint, Auto-Owners’s quarrel was with Stevens & Ricci regarding its indemnity obligation under the Policy ... [and] [i]ts dispute was thus with its insured, not the class.” Majority Op. at 399. I write separately because I am unconvinced that this approach is permissible in lighhof the anti-aggregation rule, and believe that, in reaching its conclusion, the majority obfuscates our jurisprudence in two important areas.
First, the majority’s view that the instant controversy is “unitary” is questionable as a practical matter and creates tension with our previous decisions. Plaintiff-Appellee Auto-Owners Insurance Company named Appellant Hymed Group Corporation as a defendant in its declaratory action “in the hope of attaining a binding judgment against both the insured and the injured party,” American Automobile Insurance Company v. Murray, 658 F.3d 311, 319 (3d Cir. 2011); moreover, Hymed, not Stevens & Ricci, Inc., has been defending the suit from the beginning. In other words, we are not presented with a unitary controversy between Auto-Owners and Stevens & Ricci because, ■ in reality, the presence of Hymed reflects the fact that a *411“controversy exist[s] between the insurance company and the injured [parties].” Id. at 319.
It is hard to conceive of what controversy actually exists between Auto-Owners and Stevens & Ricci given the fact that Stevens & Ricci has not so much as entered an appearance in the matter. As Auto-Owners itself states in its Brief: “Hymed and Auto-Owners are the parties currently engaged in the ‘real dispute’ that has reached this Court on appeal ... and Hymed (and the class members) are the only ones that have a financial interest in the coverage issue .... ” Appellee’s Supp. Br. at 12.
The notion that insurance coverage disputes occur only between the insurance company and its insured fits uneasily with our Article III standing decisions in which we have stressed that parties like Hymed have a significant stake. See Murray, 658 F.3d at 319 (explaining that an injured party has a “particularized interest” in an insurance coverage suit “because a determination of ... coverage would dictate its ability to receive the full benefit of the ... lawsuit”); see also Federal Kemper Ins. Co. v. Rauscher, 807 F.2d 345, 354 (3d Cir. 1986) (“Concluding that the injured party has an independent, and not a derivative right, to be heard, is not only jurispruden-tially sound, but is also realistic[.]”).
As the majority notes, the import of its decision is that parties like Hymed cannot assert federal jurisdiction in declaratory actions seeking similar relief. In my view, this retreats from our previous decisions emphasizing the particularized interest of the injured party. The majority observes that “[standing and amount-in-controversy are two distinct inquiries.” Majority Op. at 400. It misses the point. My uneasiness with the majority’s characterization of coverage disputes arises from the practical anomaly of the real party in interest losing part of its stake in the suit.
Further, I believe that the majority’s approach is essentially a run around the anti-aggregation rule. We have prohibited measuring the amount-in-controversy by the defendant’s total cost on the basis that it violates anti-aggregation principles. In Packard v. Provident National Bank, 994 F.2d 1039, 1050 (3d Cir. 1993) we stressed that “allowing the amount in controversy to be measured by the defendant’s cost would eviscerate Snyder [v. Harris, 394 U.S. 332, 89 S.Ct. 1053, 22 L.Ed.2d 319 (1969)]’s holding that the claims of class members may not be aggregated in order to meet the jurisdictional threshold,” and thus declined to do so.
The majority’s response to this reasoning is that it is not aggregating but rather assessing the amount-in-controversy as the total “value of the right being litigated” from the perspective of Auto-Owners. Majority Op. at 400. On this point, the majority invokes the plaintiffs viewpoint rule, under which the test for determining the amount-in-controversy relies solely on the value of the benefit to the plaintiff. In other words, the majority believes that because the defendant in the underlying action is now the plaintiff, the reasoning in Packard does not apply.
But the majority’s approach does not actually elude the aggregation of class members’ claims. This becomes clear if we consider courts’ treatment of the “either viewpoint” approach, under which the amount-in-controversy is based on the pecuniary result to either party that would be produced by the judgment. See 14AA Charles A. Wright et al., Federal Practice and Procedure § 3702.5 (4th ed. 2016).
Courts addressing the either-viewpoint approach have declined to adopt the rule in suits involving class actions, for the simple fact that total cost is the same as aggrega*412tion. See In re Ford Motor Co./Citibank (South Dakota), N.A., 264 F.3d 952, 958 (9th Cir. 2001). This is exactly what we explained in Packard: aggregation is not. avoided by shifting perspective or semantics — whether one calls it “total cost” or “total detriment,” or “value of the right being litigated,” if one arrives at that total through aggregation of individual claims, it violates Snyder. Concluding otherwise, we explained, disturbs long-standing anti-aggregation principles.
The Ninth Circuit Court of Appeals has made the same point. In Ford Motor Co., 264 F.3d at 958, class-action plaintiffs seeking injunctive relief invoked the “either viewpoint” rule and argued that the amount-in-controversy should be viewed as the total detriment to the defendant. The court rejected this approach, because of the “inherent conflict” between the application of the approach and the anti-aggregation rule, explaining that “ ‘total detriment’ is basically the same thing as aggregation, and ... where the equitable relief sought is but a means through which the individual claims may be satisfied, the ban on aggregation applies with equal force to the equitable as well'as the monetary relief.” Id. at 959.
Here, the relief requested is different but the implications are the same. Auto-Owners asks us to view the amount-in-controversy as its total detriment on the presumption that this eludes application of the anti-aggregation rule. But the “inherent conflict” noted in Ford does not really disappear simply by assessing the amount-in-controversy from the plaintiff-insurance company’s viewpoint. This is because in a declaratory action, the court looks to the underlying suit to determine the amount in controversy. See e.g., Jumara v. State Farm Ins. Co., 55 F.3d 873, 877 (3d Cir. 1995) (“[T]he amount in controversy in a petition to compel arbitration or appoint an arbitrator is determined by the underlying cause of action that would be arbitrated.”).1 That means, regardless of party status in the declaratory action, the plaintiffs claim in the underlying suit still provides the basis from which we calculate the amount in controversy. Put differently: merely labeling the amount as the “total amount the insurance company will owe” does not sidestep the fact that, practically speaking, we are deriving that amount by aggregating the individual claims of the class members in the underlying suit.2
The majority believes that its approach is “consistent” with Ford. Majority Op. at 400.1 disagree. Ford straightforwardly explained that it would not consider the total detriment of the defendant in reaching the amount in controversy because “total detriment” and “aggregation” are one and the same. Here, the majority concludes that total detriment and aggregation are not the same simply because the insurance company is the plaintiff in the declaratory action. Rather than being consistent with Ford, the majority skirts the relevant point of Ford through a narrow focus on party status.
Thus to calculate the amount-in-controversy, I would do what our precedents suggest: look to Hymed’s underlying suit *413to determine the amount in controversy. In doing so, it is clear to a legal certainty that the amount-in-controversy is not met. No single class member’s claim would exceed, or even come close to, the $75,000 threshold.
To be sure, the total amount potentially owed by Auto-Owners also falls short of the $75,000 threshold. The majority attempts to overcome this problem by tacking onto the amount the cost of hypothetical attorneys’ fees. I also depart from this approach. Generally, attorneys’ fees are not considered a part of the amount-in-controversy. 28 U.S.C. § 1332(a); see also 14AA Charles A. Wright et al., Federal Practice and Procedure § 3712 (4th ed. 2016). We have referenced a narrow exception to this rule when “the contracts at issue called for the payment of attorneys’ fees and costs by the party breaching the contract.” State Farm Mut. Auto Ins. Co. v. Powell, 87 F.3d 93, 98 (3d Cir. 1996). The policy between Auto-Owners and Stevens & Ricci (the “Policy”) is not an instrument of this sort.
The majority relies on Powell in concluding that attorneys’ fees are duly included here, but that case provides only apparent support. In Powell, we declined to include arbitration costs in the amount-in-controversy because the policy at issue did “not specifically impose a duty to pay on the part of the [insurance company].” Id. Similarly, the Policy contains no provision specifically imposing on Auto-Owners a duty to pay attorneys’ fees and costs. While the majority points out that the Policy imposes on Auto-Owners a general “duty to defend,” this duty only applies to suits that fall within coverage. Thus, it is not an unconditional requirement from which we could comfortably speculate as to costs that may be incurred.
The way the majority presents it, it might appear that Powell straightforwardly adopts the position it embraces. Not so. The full text of the language cited by the majority is as follows:
As an initial matter, we question the reasoning of the district court’s decision in [Nationwide Mutual Insurance Company v. Rowles by Rowles, 818 F.Supp. 852 (1993)]. In arriving at its conclusion, the Rowles court relied upon two cases, [Springstead v. Crawfordsville State Bank, 231 U.S. 541, 34 S.Ct. 195, 58 L.Ed. 354 (1913)] and [Farmers Insurance Company v. McClain, 603 F.3d 821 (10th Cir. 1979)] which held that costs and attorneys’ fees should be considered part of the amount in controversy for jurisdictional purposes when they are mandated by underlying instruments or contracts. In those two cases, however, the contracts at issue called for the payment of attorneys’ fees and costs by the party breaching the contract.
Id. at 98. What should be readily apparent is that the majority takes Powell out of context. There, we did not hold that “costs and attorneys’ fees should be considered part of the amount in controversy for jurisdiction purposes when they are mandated by underlying instruments or contracts;” we merely cited a district court case that made a conclusion to that effect. The upshot of Powell’s holding is more limited — it is, where an underlying instrument does not “specifically impose a duty to pay” fees and costs, such costs are not duly included in the amount in controversy. Id.
To be clear, I would agree that where a contract requires a breaching party to pay attorneys’ fees, those may be considered part of the amount in controversy — after all, in those contexts, “the costs [are] essentially additional damages to be assessed against the party found to have breached the instrument.” Id. This narrow exception to the general prohibition does not apply *414here. Whatever fees and costs flowed from Auto-Owners’s duty to defend in its subsequent defense of Stevens & Ricci are entirely forward-looking — and as the majority opines, “federal diversity jurisdiction is generally determined based on the circumstances prevailing at the time the suit was filed.” Majority Op. at 400 (citing Kaufman v. Allstate N.J. Ins. Co., 561 F.3d 144, 152 (3d Cir. 2009)). And these costs are not “essentially additional damages” Auto-Owners owes to Stevens & Ricci or vice versa, especially not as it relates to the declaratory judgment action. It would be a different scenario if Stevens & Ricci sued Auto-Owners for breach of the Policy’s duty to defend. But that is not the situation we are presented with here.
In sum, the inclusion of attorneys’ fees here is an unnecessary expansion of our jurisprudence for which the majority articulates no basis.3 The majority believes that those fees are an “inseparable” part of the controversy and to ignore them is to “ignore the reality of what is at stake in-this litigation.” Majority Op. at 402. But in my view, by including these fees, the majority ignores the express language of 28 U.S.C. § 1332(a) in the absence of any support from our Court.
“It is axiomatic that federal courts are courts of limited jurisdiction, and as such are under a continuing duty to satisfy themselves of their jurisdiction before proceeding to the merits of any case.” Packard, 994 F.2d at 1049 (citations omitted). Here, I am afraid the majority has “ben[t] over backwards ... to persuade itself that subject matter jurisdiction exists.” Travelers Prop. Cas. v. Good, 689 F.3d 714, 718 (7th Cir. 2012).
Thus, because of the difficult fit between the majority’s reasoning and our Article III standing jurisprudence in similar contexts, the tension between the insurance-company viewpoint approach and the anti-aggregation rule, and my reservations regarding the inclusion of the attorneys’ fees to reach the amoúnt-in-controversy, I am not satisfied of our jurisdiction over this dispute. I would dismiss the appeal on that ground and therefore respectfully dissent.
. The majority calls this approach unprecedented. But we have clearly suggested as much in Jumara. Further, the proposition that we look to the underlying suit to determine the amount in controversy is apparent as a matter of common sense.
. This is so even if we adopt the fiction that because declaratory relief is sought, the controversy becomes unitary, only involving the insurer and its insured. Even viewing this suit in this way, we do not derive the amount of controversy out of thin air — rather, we must look to the value of damages sought in the underlying suit.
. The majority cites an influential treatise for the proposition that the law in this area is "quite settled.” Majority Op. at 401-02 n.18. However, one treatise does not law make. And, a closer look at the cases collected in that treatise demonstrates that what is "quite settled” is that where an underlying contract contains a fee-shifting provision, attorneys' fees may be included in the amount-in-controversy, a point I do not dispute. See 14AA Charles Alan Wright et al., Federal Practice & Procedure § 3712 (4th ed. 2016). What is less settled, and has received less circuit attention, is the context we are presented with here— with only one circuit sharing the majority's view and other circuits employing a similar approach only where the relevant defense or indemnification occurs before the declaratory suit is brought. See Stonewall Ins. Co. v. Lopez, 544 F.2d 198, 199 (5th Cir. 1976) (per curiam) (cost of insurer's prospective defense of insured provides basis for including attorneys’ fees); S. Ariz. York Refrigeration Co. v. Bush Mfg. Co., 331 F.2d 1, 18 (9th Cir. 1964) (attorneys' fees included when defense has already occurred and indemnitor refused to defend); Farmers Ins. Co. v. McClain, 603 F.2d 821, 823 & 823 n.3 (10th Cir. 1979) (attorneys’ fees included when defense has already occurred). Far from settled then, the circuit law addressing this particular question is limited and inconclusive.