NOT PRECEDENTIAL

                UNITED STATES COURT OF APPEALS
                     FOR THE THIRD CIRCUIT
                         _______________

                             No. 10-2343
                           _______________

                BASSEM KANDIL; FLORA KANDIL
              SAMEH A. ABOELATA; HALLA KANDIL,

                                               Appellants

                                    v.

               GARY YURKOVIC, POLICE OFFICER;
            ANTHONY MARK ABODE, POLICE OFFICER;
WILLIAM C. OELS, III, POLICE OFFICER; WILLIAM OELS, JR., SERGEANT;
            EDWARD T. BOBADILLA, POLICE OFFICER;
           CHIEF OF POLICE; CITY OF NEW BRUNSWICK;
         CITY OF NEW BRUNSWICK POLICE DEPARTMENT;
           MIDDLESEX COUNTY PROSECUTOR’S OFFICE;
     MIDDLESEX COUNTY CORRECTIONAL FACILITY, JOHN DOE
   SUPERVISING OFFICERS 1-10; JOHN DOES 1-10; ABC CORPS, 1-10

                           _______________

             On Appeal from the United States District Court
                       For the District of New Jersey
                  (D.C. Civil Action No. 2-06-cv-04701)
             District Judge: Honorable Dennis M. Cavanaugh
                             _______________

               Submitted Under Third Circuit LAR 34.1(a)
                          September 20, 2011
                          _______________

       Before: AMBRO, CHAGARES, and ALDISERT, Circuit Judges

                    (Opinion filed: October 17, 2011 )
                                     _______________

                                        OPINION
                                     _______________

AMBRO, Circuit Judge

       Appellants Bassem Kandil and his family members appeal from a summary

judgment entered against them by the District Court. Kandil argues that it improperly

withheld jurisdiction over his case and that, as a matter of public policy, the release-

dismissal agreement he signed should not be enforced against him. For the reasons that

follow, we hold that federal jurisdiction exists over this case and that there are genuine

issues of material fact as to the enforceability of the agreement. Accordingly, we vacate

the District Court’s judgment and remand.

I.     Background

       New Brunswick, New Jersey police officers arrested Kandil in October 2004. He

was charged with aggravated assault, resisting arrest, and disarming a police officer. A

Middlesex County grand jury subsequently indicted Kandil on these charges. In the

ensuing months, both the State and Kandil conducted discovery. The State also offered

Kandil a plea agreement, which he declined. Throughout these proceedings and in this

appeal, the same counsel has represented Kandil.

       In October 2005, Judge Frederick DeVesa of the Superior Court of New Jersey

held a status conference. At that conference, Assistant Prosecutor Marcia Silva renewed

the State’s plea offer. However, the State declined to enroll Kandil in the Pretrial

Intervention program (―PTI‖), which would have suspended the charges against him


                                              2
during two years of supervised release, after which the charges would be dropped. Judge

DeVesa then invited the parties’ counsel into his chambers for an off-the-record

negotiation. During that negotiation, he learned that Kandil had filed a Notice of Tort

Claim under the New Jersey Tort Claims Act. The parties’ counsel then reached a

compromise: Kandil could enroll in the PTI program, but in exchange he would waive all

civil claims against State entities and officials.

       Kandil applied for admission to the PTI program the very day of the status

conference. However, ten days later, the Probation Department denied Kandil’s

application. To justify its denial, the Department cited the violent nature of Kandil’s

alleged offense, the public need for prosecution, and the gravity of what might have

occurred had Kandil wrested control of the officer’s weapon. On learning of this denial,

Assistant Prosecutor Silva (who had promised Kandil’s admission to the program) sought

an override of the denial from First Assistant Prosecutor Julia McClure. McClure granted

the override, provided that Kandil attend anger management classes and, again, waive all

civil claims.

       In November 2005, the parties returned to Judge DeVesa’s courtroom for a PTI

enrollment hearing. There, the State formally enrolled Kandil in the PTI program. Judge

DeVesa also specifically questioned Kandil and his counsel about Kandil’s waiver of any

civil claims. Both Kandil and his counsel affirmed that they understood the implications

of that waiver. The parties stipulated that they would sign a written release-dismissal

agreement waiving civil claims by December 2, 2005.



                                               3
       In September 2006, Kandil filed this civil case in the District Court, having never

signed a release-dismissal agreement waiving those claims. Kandil’s claims against the

police officers include common-law tort claims and claims under 42 U.S.C. § 1983. The

Probation Department, observing that a waiver of precisely such claims was a condition

of Kandil’s enrollment in the PTI program, filed a Notice of Motion for Termination in

February 2007. Thereafter, Kandil signed a release-dismissal agreement releasing all

civil claims arising out of his arrest, including the pending matter in the District Court.

This agreement satisfied the Probation Department, and Kandil completed the PTI

program in November 2007. Therefore, the underlying criminal charges against Kandil

were dismissed.

       The defendants, citing Kandil’s written release of his claims against them, moved

in the District Court for summary judgment. In granting the motion, the Court held that it

lacked jurisdiction under the Rooker-Feldman doctrine (discussed below) and, in the

alternative, that Kandil’s release-dismissal agreement is enforceable against him as a

matter of public policy. Kandil now appeals to our Court both grounds of that holding.

For the reasons that follow, we agree with Kandil at least that there is a genuine issue of

material fact and that summary judgment was thus inappropriate. We therefore vacate

the District Court’s grant of summary judgment and remand.

II.    Discussion

       The District Court had jurisdiction under 28 U.S.C. § 1331. We have jurisdiction

pursuant to 28 U.S.C. § 1291. Our review in the summary judgment context is plenary.

Thus, ―we are required to apply the same test that the district court should have utilized

                                              4
initially.‖ Jackson v. Danberg, 

, 215 (3d Cir. 2010) (quotation marks and

citation omitted). Summary judgment is appropriate only when there are no genuine

issues of material fact, drawing all justifiable inferences in favor of the non-movant.

Anderson v. Liberty Lobby, Inc., 

, 248, 255 (1986).

       A.     Rooker-Feldman Doctrine

       The District Court held that it lacked jurisdiction over this case because of the

Rooker-Feldman doctrine. We disagree.

       That doctrine, which takes its name from two Supreme Court cases, generally

withholds jurisdiction of federal courts (save the United States Supreme Court) over state

judgments, as they are more appropriately appealed within the state judiciary. See D.C.

Court of Appeals v. Feldman, 

(1983); Rooker v. Fid. Trust Co., 

(1923). It is narrow in scope. The Supreme Court has explained that it ―is confined

to cases of the kind from which the doctrine acquired its name: cases brought by state-

court losers complaining of injuries caused by state-court judgments rendered before the

district court proceedings commenced and inviting district court review and rejection of

those judgments.‖ Exxon Mobil Corp. v. Saudi Basic Indus. Corp., 

, 284

(2005). We have distilled Exxon’s holding into the following four-part test, each part of

which must be satisfied: ―(1) the federal plaintiff lost in state court; (2) the plaintiff

complains of injuries caused by the state-court judgments; (3) those judgments were

rendered before the federal suit was filed; and (4) the plaintiff is inviting the district court

to review and reject the state judgments.‖ Great W. Mining & Mineral Co. v. Fox

Rothschild LLP, 

, 166 (3d Cir. 2010) (quotation marks and citation omitted).

                                                5
       This case does not meet those criteria.1 Kandil did not ―lose‖ in state court. The

state court order at issue is Judge DeVesa’s Order of Postponement (of Kandil’s criminal

charges) that memorializes the parties’ compromise: the State would suspend its charges

against Kandil, which it did, and Kandil would waive any civil charges against State

actors, which he (eventually) did.

       More importantly, the injuries that Kandil asserts in this suit and on appeal were

not ―caused by the state-court judgment.‖ The core of Kandil’s complaint is his tort and

§ 1983 claims against the police officers. Those claims existed before any state court

case; the state-court judgment in no sense ―caused‖ those alleged injuries. See Great W.

(citing examples of injuries that do and do not meet this

criterion and offering the ―timing of the injury‖ as a ―useful guidepost‖ for resolving the

question); Turner v. Crawford Square Apts. III, L.P., 

, 547 (3d Cir. 2006)

(―Turner’s complaint raised federal claims, grounded on the FHA, not caused by the

state-court judgment but instead attributable to defendants’ alleged FHA violations that

preceded the state-court judgment.‖) (emphasis in original).

       Kandil does not technically ―complain‖ that his waiver caused injury, as waiver is

asserted against him as an affirmative defense. Nonetheless, his waiver is in effect the

main issue before us. He argues that his release-dismissal agreement is unenforceable,


1
 The District Court relied on one of our pre-Exxon holdings to conclude that Rooker-
Feldman applies. However, Exxon is read as a narrowing of our Court’s earlier Rooker-
Feldman jurisprudence. Thus, we have noted that ―caution is now appropriate in relying
on our pre-Exxon formulation of the Rooker-Feldman doctrine.‖ Gary v. Braddock
Cemetery, 

, 200 n.5 (3d Cir. 2008).

                                             6
which is a question of contract law and public policy. That a state-court order recited that

the waiver was in exchange for entry into the PTI program is, for Rooker-Feldman

analysis, of no import. In this respect, Kandil’s position is like that of the plaintiff in

Skinner v. Switzer, 

(2011). He lost in state court, then sued in federal

court to challenge the constitutionality of the state laws under which he was convicted.

The Court held that Rooker-Feldman did not bar his claim. 

Similarly here,

Kandil challenges the legality of the contract that would govern the state court decision,

not a state court decision itself.

       We conclude that the Rooker-Feldman doctrine does not preclude jurisdiction in

this case. We must thus consider the District Court’s alternative basis for its holding –

that Kandil’s waiver of civil claims is enforceable and requires judgment against him.

       B.      Release-Dismissal Agreement

       The District Court held that even if it had jurisdiction, Kandil’s release-dismissal

agreement must be enforced against him, barring this suit. We hold that there are

genuine issues of material fact, and thus we vacate and remand.

       Our law places few limits on the substance of contractual bargains. In particular,

parties may give up important legal rights to gain a benefit they seek, provided that their

agreement is knowing and voluntary. See, e.g., Brady v. United States, 

(1970) (holding that plea agreements are enforceable). However, the Supreme Court has

constrained the enforceability of agreements that waive civil suits in exchange for release

from prosecution, such as the release-dismissal agreement in this case. See Town of

Newton v. Rumery, 

(1987).

                                               7
       Rumery establishes two requirements for a release-dismissal agreement to be

enforceable. First, the agreement must be 

. We answer the

question of voluntariness with reference to ―the knowledge and experience of the

defendant, the severity of the criminal charges, whether a legitimate criminal justice

objective supports the release, whether the defendant was represented by counsel, and

whether the execution of the release was judicially supervised.‖ Livingstone v. North

Belle Vernon Borough, 

, 1210 (3d Cir. 1993) (en banc) (―Livingstone I‖)

(quotation marks and citation omitted). Second, enforcement of the agreement must be in

the public interest, an inquiry that emphasizes avoiding prosecutorial misconduct.

. The burden is on the civil-case defendants to ―show that

upon balance the public interest favors enforcement.‖ Cain v. Darby Borough, 

, 381 (3d Cir. 1993) (en banc). In this context, we ask whether the facts known to the

prosecutor at the time of the agreement support the proffered public interest reason for

the agreement and whether the proffered reason is the prosecutor’s actual reason for

seeking the agreement. Livingstone v. North Belle Vernon Borough, 

, 527

(3d Cir. 1996) (―Livingstone II‖).

       Kandil concedes that the release-dismissal agreement was knowing and voluntary.

See Kandil Br. at 8, 28. The Rumery Court found it important that the plaintiff was a

businessman. Kandil is an auto salesman and finance manager. The Rumery Court found

it important that the plaintiff was not in jail when he made the agreement. Kandil was not

in jail. The Rumery Court found it important that an experienced criminal lawyer

represented the plaintiff. Kandil has been represented throughout this case by a lawyer

                                             8
who was an assistant prosecutor for nine years. The Rumery Court found it important

that the plaintiff’s lawyer had drafted the agreement. Kandil’s lawyer drafted the

agreement. The Rumery Court found it important that the plaintiff had three days to

consider the agreement. Kandil had much longer between Judge DeVesa’s suggestion of

the agreement on October 18 and Kandil’s decision to enroll in the PTI program, thereby

committing to the agreement on November 9. We therefore agree that Kandil’s

agreement was voluntary.

       The second question, whether enforcement of Kandil’s agreement is in the public

interest, is more challenging. The officials seeking the agreement’s enforcement must

present a legitimate public-interest reason why they sought the waiver. In this context,

the ―public interest‖ has included subjecting the agreement to judicial supervision,

avoiding stress on defendants and key witnesses, avoiding the costs of prosecution, and

avoiding frivolous civil lawsuits. See Livingstone 

. We then subject

this reason to an ―objective‖ and a ―subjective‖ analysis. The ―objective‖ analysis asks

whether the facts known to the prosecutor at the time of the agreement suffice to establish

that it was in the public interest. The ―subjective‖ analysis asks whether the reason

proffered was in fact the reason that the prosecutor sought the agreement.

       Defendants advance three reasons why they asked Kandil to waive his civil claims

in exchange for admission to the PTI program. First, they contend that the prosecutors,

Silva and McClure, admitted Kandil to the program for legitimate reasons that included

his lack of criminal history and the fact that he merely unholstered the officer’s weapon

rather than seizing it. Second, they argue that because Judge DeVesa was the first to

                                             9
suggest the release-dismissal agreement, the prosecutors did not seek the agreement to

cover their own misconduct. Third, Kandil’s civil suits, they offer, would be marginal

and judicial resources are better used elsewhere. They base this argument on internal

investigations by the police department absolving the officers of bias and excessive force.

       Our ―objective‖ analysis of these three reasons raises questions about whether they

suffice to establish that the release-dismissal agreement was in the public interest. The

first reason might illuminate why the prosecutors enrolled Kandil in the PTI program, but

it says little about why waiver of civil suits was a precondition. Indeed, the prosecutors

knew those facts—that Kandil merely unholstered the officer’s weapon, that he had no

criminal record, that he was employed and married—before the October 2005 status

conference. Yet at that status conference they refused to enroll Kandil in the PTI

program. Only after the status conference, once the waiver of civil claims was on the

table, did the prosecutors change their mind.

       The second reason, that Judge DeVesa suggested the agreement, is far from the

kind of judicial supervision that would put the agreement in the public interest. The

parties dispute that Judge DeVesa, as opposed to the prosecution, suggested the

agreement, as their conference was off the record. Even if Judge DeVesa did suggest the

agreement, he hardly supervised it. At the PTI enrollment hearing, he noted that ―I see

here in the proposed order for the Court’s signature that the defendant as a condition of

Pretrial Intervention intends to execute some type of a [hold] harmless agreement against

[the] officers . . . .‖ App. 506. This remark does not show close scrutiny or an awareness

of the merits of Kandil’s civil claims.

                                            10
       The third proffered reason, that Kandil’s civil suits appeared marginal or frivolous,

requires closer analysis. In Livingstone II, we wrote that unless ―unusually strong public

interests‖ support the other reasons, prosecutors must conduct an ―individualized

analysis‖ of the plaintiff’s potential civil claims before seeking a release-dismissal

& n.16. In this case, the police department’s internal

investigations are the kind of case-specific inquiry that Livingstone II requires. However,

as in Livingstone II, the District Court here did not consider whether these investigations

were enough to support the conclusion that Kandil’s civil claims were marginal or

frivolous. More importantly, as in Livingstone II, ―[i]t is not now clear whether [the

prosecutors] considered the marginal or frivolous nature of [Kandil’s] claims‖ in deciding

to seek a waiver. 

n.17. Rather, the only reasons they offered at their

depositions were that Judge DeVesa had suggested the waiver and that Kandil had agreed

to it. App. 411 (McClure), 499 (Silva). Thus, as in Livingstone II, we remand ―so that

the parties can address the question of whether [Kandil’s] civil claims were regarded—

and, if so, whether they were properly regarded—by the prosecuting attorney as marginal

or 

.

       Our ―subjective‖ analysis of appellees’ proffered reasons raises still more

questions. In September 2005, before anyone suggested that Kandil waive his civil

claims, Kandil’s counsel submitted discovery evidence of police misconduct to the

prosecutors. That evidence, consisting largely of interviews with witnesses and an

interrogatory answered by Officer Yurkovic, suggests an alternate motive for the release-

dismissal agreement. It indicates that Yurkovic was having a ―casual sexual relationship‖

                                             11
with a woman named Pamela, that Kandil and his friends were conversing with Pamela

and her friend on the night of the incident, that Yurkovic arrived and demanded Kandil’s

driver’s license on observing him with Pamela, and that Yurkovic may have thereafter

cursed, mocked Kandil’s home town of Flemington, and grabbed Kandil. App. 600-09,

635-36.

       Defendants do not challenge this evidence directly. Rather, they note that no

discovery about these events has been undertaken in this civil case, as opposed to

Kandil’s criminal case. Appellee Br. at 72. More importantly, they emphasize that

prosecutors reviewed Kandil’s discovery evidence and concluded that they should

proceed with their prosecution in spite of it. 

69, 73. However, the record on

which defendants found their assertion suggests that the official in the prosecutor’s office

responsible for allegations like Kandil’s may have only ―flipped through‖ the discovery

materials before concluding that ―there doesn’t seem to be a problem here.‖ App. 819.

Regardless of the thoroughness of the inquiry into the discovery, the record establishes

that the prosecutors were aware of this evidence of misconduct before they sought the

release-dismissal agreement. As in Livingstone II, that awareness ―lends credence to the

inference‖ that the prosecutors’ subjective motivation for pursuing the waiver agreement

was not purely in the public interest. 

.

       The record contains other evidence tending to make the prosecutors’ subjective

motivation a genuine issue of material fact. In early November 2005, about a month and

a half after Abramowitz determined that Kandil’s discovery evidence did not present a

problem, Assistant Prosecutor Silva spoke to First Assistant Prosecutor McClure. In that

                                             12
discussion about enrolling Kandil in PTI, she raised ―this issue of the two girls [Pamela

and her friend] and were they there and weren’t they there and that kind of thing.‖ App.

498. ―This issue‖ may thus have been relevant to the prosecutors’ decision to put Kandil

in the PTI program in exchange for his release-dismissal agreement. The same

deposition testimony reveals that this prosecutor’s office had no history of requiring

release-dismissal agreements to enroll in the PTI program. App. 496. That one was

required in this case, and only in this case, could suggest that they paid additional

attention to their decision.

       On this important question of the agreement’s public-interest merits, the District

Court is silent. It properly states the law and recognizes that the question demands an

objective and a subjective analysis. However, in holding that Rumery does not bar

enforcement of the agreement, the Court observes only that Kandil received the benefit of

his bargain and that he was represented by counsel.

       We therefore vacate and remand the judgment of the District Court so that it might

assess in the first instance whether, under Livingstone II, enforcement of Kandil’s

release-dismissal agreement is in the public interest. On remand, the parties may wish to

address more thoroughly the prosecutors’ motives for seeking the agreement and the

relationship of Kandil’s discovery evidence to that decision. Evidence such as the

deposition testimony of First Assistant Prosecutor McClure, which was provided to us

only in part, may be pertinent to making this assessment.




                                             13