Five named plaintiffs, members of Indian tribes and present or past beneficiaries of Individual Indian Money (“IIM”) accounts, filed a class action in district court in 1996, alleging that the defendants — the Secretaries of the Interior and the Treasury, and the Assistant Secretary of the Interior for Indian Affairs — had “grossly mismanaged” those accounts. The bulk of the funds in the accounts are the proceeds of various transactions in land allotted to individual Indians under the General Allotment Act of 1887, known as the “Dawes Act,” ch. 119, 24 Stat. 388 (codified as amended at 25 U.S.C. § 331 et seq. (§§ 331-333 repealed 2000)). The money-producing transactions in question evidently involved such matters as sales of timber and leases of rights to grazing, farming, or extraction of oil, gas, or other minerals. Complaint, ¶¶2, 3, 5, 7-11, 17. See also Cobell v. Babbitt, 91 F.Supp.2d 1, 9-12 (D.D.C.1999) (“Cobell V”). (The accounts also contain funds from a variety of other sources, see 25 C.F.R. § 115.702, but the allotment land transactions apparently predominate.)
Plaintiffs’ suit draws significantly on Congress’s findings of hopelessly inept management of the IIM accounts and its action to remedy the resulting chaos. A 1992 Congressional report, Misplaced *94Trust: The Bureau of Indian Affairs’ Mismanagement of the Indian Trust Fund, H.R.Rep. No. 102-499 (1992), catalogued Interior’s “dismal history of inaction and incompetence,” id. at 5, and concluded that the agency had “repeatedly failed to take resolute corrective action to reform its longstanding financial management problems,” id. at 3. In 1994 Congress moved from findings to legislation, passing the Indian Trust Fund Management Reform Act, Pub. L. No. 103-412, 108 Stat. 4239 (codified as amended at 25 U.S.C. § 162a et seq. & § 4001 et seq.) (the “1994 Act”). The 1994 Act imposed a variety of duties on the Secretary of the Interior, most of them relating directly to trust funds such as the IIM accounts. See, e.g., 25 U.S.C. § 162a(d).
Even apart from the 1994 Act, the IIM funds have quite a different legal status from the allotment land itself. Section 5 of the Dawes Act nominally made the United States trustee of those lands, but did so solely in order to limit alienation by Indians and to assure immunity of the lands from state taxation. See United States v. Mitchell, 445 U.S. 535, 540-44, 100 S.Ct. 1349, 1352-54, 63 L.Ed.2d 607 (1980) (“Mitchell I”). It gave the Indian beneficiaries the right to possess and manage the lands except insofar as alienation was involved. Id. at 542-46, 100 S.Ct. at 1353-56. See also United States v. Navajo Nation, 537 U.S. 488, 504, 123 S.Ct. 1079, 1090, 155 L.Ed.2d 60 (2003) (describing Mitchell I and applying its principles to certain unallotted lands). Accordingly, the Supreme Court held in Mitchell I that the Dawes Act did not, alone, establish a fiduciary duty on the part of the United States to manage the allotted lands. 445 U.S. at 544, 546, 100 S.Ct. at 1354, 1355-56. In contrast, the IIM funds are by statute under the full control of the United States, to be invested for the benefit of individual Indians in public debt of the United States or deposited in banks. See 25 U.S.C. §§ 161a(b), 162a(a).
As the label Cobell V suggests, this litigation has generated many legal opinions, including three of this court. In Cobell v. Norton, 240 F.3d 1081 (D.C.Cir.2001) (“Cobell VI”), we affirmed the district court’s holding that the officials had breached their fiduciary duties and remanded for further proceedings. In Cobell v. Norton, 334 F.3d 1128 (D.C.Cir.2003) (“Cobell VIII”), we vacated a contempt citation of successor defendants Interior Secretary Gale Norton and Assistant Secretary of Indian Affairs Neal McCaleb, and reversed the district court’s appointment of a court monitor. And finally, in Cobell v. Norton, 391 F.3d 251 (D.C.Cir.2004), we vacated an order of the district court directing Interi- or to disconnect its computers from the Internet pending a security determination, excepting only certain essential systems and ones that would not provide access to Indian trust data. Those opinions, as well as the many opinions of the district court, provide an array of background data.
Here we address a district court injunction issued September 25, 2003. Cobell v. Norton, 283 F.Supp.2d 66 (D.D.C.2003) (“Cobell X”). The decree, see id. at 287-95, imposes obligations on the defendants in two main categories. Duties related to “Historical Accounting” are intended to unravel the tangle resulting from past accounting failures, see id. at 70-211; those related to “Fixing the System” are intended to compel the issuance of a plan for future trust administration as a whole, see id. at 239-87. To assure fulfillment of both sets of duties, the court appointed a court monitor to oversee compliance and said it would retain jurisdiction until December 31, 2009. These two different sets of commands raise quite different issues.
“Historical Accounting,” we find, is governed by Pub. L. No. 108-108, a provision *95adopted after the district court opinion issued, which radically changes the underlying substantive law and removes the legal basis for the historical accounting elements of the injunction. We therefore vacate those elements.
The core of “Fixing the System,” by contrast, requires the Interior defendants to produce a “plan” that would fix the IIM trust management system, and requires the Interior defendants to explain how the Department will comply with various constraints or objectives identified by the court, such as sixteen specific common law trust duties and tribal law. Although we agree that Interior is subject to many of the common law trust duties identified by the court, we find that much of the “Fixing the System” injunction exceeds the court’s remedial discretion because the court failed to ground it in the defendants’ statutory trust duties and in specific findings that Interior breached those duties. Aside from the requirement that Interior complete its so-called “To-Be Plan,” as promised in its Comprehensive Plan, we thus vacate the district court’s injunction and remand for further proceedings consistent with this opinion.
Historical Accounting
In Cobell VI we ruled that the 1994 Act, 25 U.S.C. § 4011(a), conferred a right on IIM beneficiaries to “a complete historical accounting of trust fund assets,” explaining that “ ‘[a]ll funds’ [as used in that provision] means all funds, irrespective of when they were deposited (or at least so long as they were deposited after the Act of June 24, 1938).” 240 F.3d at 1102. In Cobell X the district court ruled that Interior must account for all funds deposited since 1887 and issued rules permitting some accounting methods and prohibiting others — e.g., rejecting any use of statistical sampling. Cobell X, 283 F.Supp.2d at 288-90.
Defendants raise a variety of objections to the district court’s historical accounting order, but the objection based on Pub.L. No. 108-108 trumps the others. Adopted November 10, 2003, less than two months after the issuance of Cobell X, Pub.L. No. 108-108 .appropriates funds and provides as follows:
For the operation of trust programs for Indians by direct expenditure, contracts, cooperative agreements, compacts, and grants, $189,641,000, to remain available until expended: Provided, That of the amounts available under this heading not to exceed $45,000,000 shall be available for records collection and indexing, imaging and coding, accounting for per capita and judgment accounts, accounting for tribal accounts, reviewing and distributing funds from special deposit accounts, and program management of the Office of Historical Trust Accounting, including litigation support: Provided further, That nothing in the American Indian Trust Management Reform Act of 1994, Public Law 103-412, or in any other statute, and no principle of common law, shall be construed or applied to require the Department of the Interior to commence or continue historical accounting activities with respect to the Individual Indian Money Trust until the earlier of the following shall have occurred: (a) Congress shall have amended the American Indian Trust Management Reform Act of 1994 to delineate the specific historical accounting obligations of the Department of the Interior with respect to the Individual Indian Money Trust; or (b) December 31, 2004.
Pub. L. No. 108-108. A later sentence of the same section provides that the statute of limitations will not begin to run on any claim for losses or mismanagement of trust *96funds “until the affected tribe or individual Indian has been furnished with an accounting of such funds from which the beneficiary can determine whether there has been a loss.” Id.
Thus Pub. L. No. 108-108 appears to give Interior temporary relief from any common law or statutory duty to engage in historical accounting for the IIM accounts. The provision’s legislative history makes clear that Congress passed it in response to Cobell X, to clarify Congress’s determination that Interior should not be obliged to perform the kind of historical accounting the district court required. The conference committee explained that “[i]nitial estimates indicate that the accounting ordered by the Court would cost between $6 billion and $12 billion .... ” H.R. Conf. Rep. 108-330, at 117. The committee “rejeet[ed] the notion that in passing the American Indian Trust Management Reform Act of 1994 Congress had any intention of ordering an accounting on the scale of that which has now been ordered by the Court. Such an expansive and expensive undertaking would certainly have been judged to be a poor use of Federal and trust resources.” Id. at 118. “Indian country would be better served by a settlement of this litigation than the expenditure of billions of dollars on an accounting.” Id. at 117. Congress thus gave itself until the end of 2004 to come up with a legislative solution. See id. at 118.
In addition, individual legislators said in effect that the disparity between the costs of the judicially ordered accounting, and the value of the funds to be accounted for, rendered the ordered accounting, as one senator put it, “nuts”: “If this is a $13 billion fund, or somewhere in the neighborhood of $13 billion, would the Native Americans want us to begin a process in which we spend up to $9 billion to hire accountants and financial folks and others to sift through these accounts? I think that is just nuts. That doesn’t make any sense at all to anybody.” 149 Cong. Rec. at S13.786 (2003) (statement of Sen. Dorgan). See also id. at S13,785 (statement of Sen. Burns) (“If there is one thing with which everybody involved in this issue seems to agree, it is that we should not spend that kind of money on an incredibly cumbersome accounting that will do almost nothing to benefit the Indian people.”).
Plaintiffs make a vague claim that we should simply disregard Pub. L. No. 108-108, allowing the district court to address its effect in the first instance. But apart from an allusion to the possibility of considering it in conjunction with post-decree developments, they offer no reason overcoming the usual principle that a court is to apply the law in effect at the time the court rules. See Landgraf v. USI Film Products, 511 U.S. 244, 264, 114 S.Ct. 1483, 1496-97, 128 L.Ed.2d 229 (1994). As the provision deprives the decree’s “historical accounting” mandates of any legal basis, it is hard to see how post-decree developments could affect the matter. As a fallback position, plaintiffs argue that the law violates separation of powers principles and the takings and due process provisions of the Fifth Amendment. We reject both claims.
First, plaintiffs assert that Pub. L. No. 108-108 amounts to a “legislative stay” of a final judicial judgment. They cite language in Hayburn’s Case, 2 U.S. (2 Dall.) 409 (1792), to the effect that Article III judicial decisions cannot “be liable to a revision, or even suspension, by the legislature.” Id. at 413 (emphasis added) (quoting decision of the circuit court for the district of North Carolina, consisting of Iredell, Justice, and Sitgreaves, district judge). In Plaut v. Spendthrift Farm, Inc., 514 U.S. 211, 115 S.Ct. 1447, 131 L.Ed.2d 328 (1995), the Court explained that Hayburn’s Case “stands for the principle that Congress cannot vest review of *97the decisions of Article III courts in officials of the Executive Branch,” id. at 218, 115 S.Ct. at 1452-53, and held that Congress could not require a federal court to reopen a completed case for money damages, id. at 240, 115 S.Ct. at 1463. But the Court also said that an appellate court must apply any law enacted after the judgment under review and clearly intended to have retroactive effect. See id. at 226, 115 S.Ct. at 1456-57.
Even more critical is the distinction between statutes that in effect reverse final judgments in suits for money damages, as in Plaut, and ones that alter the substantive obligations of parties subject to ongoing duties under an injunction, as in Pennsylvania v. Wheeling & Belmont Bridge Co., 59 U.S. 421, 18 How. 421, 15 L.Ed. 435 (1855). Indeed, Plant explicitly distinguished the latter. See 514 U.S. at 232, 115 S.Ct. at 1459-60. In Wheeling Bridge a comb had entered a decree requiring removal of a bridge pursuant to a statute rendering it unlawful. Congress then amended the law to legalize the bridge. The Court held that because the act of Congress modified the law “so that the bridge is no longer an unlawful obstruction, it is quite plain the decree of the court cannot be enforced.” 59 U.S. at 432. For purposes of the rule limiting congressional reversal of final judgments, an injunction is not “final.” As we said in National Coalition To Save Our Mall v. Norton, 269 F.3d 1092 (D.C.Cir.2001), applying Wheeling Bridge, “[Although an injunction may be a final judgment for purposes of appeal, it is not the last word of the judicial department because any provision of prospective relief is subject to the continuing supervisory jurisdiction of the court, and therefore may be altered according to subsequent changes in the law.” Id. at 1096-97 (quoting Miller v. French, 530 U.S. 327, 347, 120 S.Ct. 2246, 2258-59, 147 L.Ed.2d 326 (2000))(internal quotation marks omitted).
At oral argument plaintiffs seemed more to stress the idea that Pub. L. No. 108-108, rather than changing the substantive law, directed the courts how to interpret or apply pre-existing law. In Save Our Mall we assumed that under United States v. Klein, 80 U.S. (13 Wall.) 128, 20 L.Ed. 519 (1871), such an interpretive direction would invade the powers of the judicial branch. 269 F.3d at 1097. Here as there, however, we do not read the statutory language as such a directive. Some of the phrasing — especially the statement that nothing in the 1994 Act or any statute or the common law “shall be construed or applied to require the Department of the Interior to commence or continue historical accounting activities” (emphasis added) — might be said to support such a reading. But “as between two possible interpretations of a statute, by one of which it would be unconstitutional and by the other valid, our plain duty is to adopt that which will save the act.” NLRB v. Jones & Laughlin Steel Corp., 301 U.S. 1, 30, 57 S.Ct. 615, 620-21, 81 L.Ed. 893 (1937).
We believe Pub. L. No. 108-108 is most plausibly read simply to say that the Department of Interior shall not, under any statute or common law principle, be required to engage in historical accounting in the specified period, i.e., all statutes and common law rules requiring any such accounting are temporarily and partially repealed or modified. Compare Robertson v. Seattle Audubon Soc., 503 U.S. 429, 112 S.Ct. 1407, 118 L.Ed.2d 73 (1992) (rejecting claim that statute should be construed as mandate of judicial findings under unchanged substantive law rather than as a change in the law). Indeed, the Supreme Court has interpreted very similar wording — that “nothing ... shall be construed” to allow — as simply repealing prior legislation to the contrary. See Carroll v. United States, 354 U.S. 394, 408-415, 77 S.Ct. *981332, 1340-1344, 1 L.Ed.2d 1442 (1957); see also Total TV v. Palmer Communications, Inc., 69 F.3d 298, 302-03 (9th Cir.1995).
Finding neither an effort to mandate a particular interpretation of the substantive law nor an impermissible legislative modification of a final judgment, we reject plaintiffs’ separation of powers theories.
Second, plaintiffs say that Pub. L. No. 108-108 is an unconstitutional deprivation of property, in violation of the due process and takings clauses of the Fifth Amendment. The claim is obscure, as plaintiffs do not explicitly identify the property right that they believe enforcement of Pub. L. No. 108-108 would take. They do, however, mention the right to “interest earned on trust accounts,” if only in a parenthetical to a case citation. Plaintiffs’ Brief at 53.
But we see no reason to think Pub. L. No. 108-108 will affect plaintiffs’ entitlement to interest. As trust income beneficiaries are typically entitled to income from trust assets for the entire period of their entitlement to income, and for imputed yields for any period of delay in paying over income or principal, see G.G. Bogert & G.T. Bogert, Law of Trusts and Trustees § 814, pp. 321-25 (rev.2d ed. 1981), we do not see — and plaintiffs make no effort to explain — how the accounting delay allowed by Pub. L. No. 108-108 could deprive them of interest or any comparable returns.
Plaintiffs’ references to temporary takings suggest that they regard a delay in the accounting itself as a taking. But the accounting is a purely instrumental right— a way of finding out the size of their claims. If the moratorium imposed by Pub. L. No. 108-108 actually delays conclusion of the accounting (which it may not, as Congress may provide a simpler scheme than the district court’s, while nonetheless assuring that each individual receives his due or more), the ordinary trust principles referred to above will automatically give the plaintiffs compensation for the delay.
Accordingly we find no constitutional obstacle to enforcement of Pub. L. No. 108-108 as written.
* :|= * * * *
In Pub. L. No. 108-108 Congress in effect gave itself until December 31, 2004 “to develop a comprehensive legislative solution to what has become an intractable problem.” H.R. Conf. Rep. 108-330, at 118. Absent Congressional action by that date, obviously Pub. L. No. 108-108 will cease to bar the historical accounting provisions of the injunction. We do not address the issues that would be relevant if the district court then reissued those provisions. At the present time, however, they are without legal basis.
Fixing the System
Although the defendants argue that Pub. L. No. 108-108 “deprives the injunction of any arguable legal basis” (Defendants’ Br. at 40), the statute suspends only “historical accounting activities.” Because certain portions of the district court’s injunction are at least conceptually separable from the historical accounting duty, we must address these aspects of the order on the merits.
What we will call Part III(IV) of the injunction (mislabeled Part III by the district court because there is already a Part III), “Compliance with Fiduciary Obligations,” is primarily an order that Interi- or complete its To-Be Plan within 90 days. Cobell X, 283 F.Supp.2d at 290-91. The To-Be Plan, which Interior sketched out broadly in its Comprehensive Plan, is intended “to provide a comprehensive statement of the manner in which trust management will be conducted after Interior’s proposed internal changes.” Id. at 250. Given that the Comprehensive Plan only *99described Interior’s intention to create the To-Be Plan, the court said that the Comprehensive Plan was “really only a plan to make a plan.” Id. at 284. Part III(IV) also orders the Interior defendants to implement the Comprehensive Plan (including the To-Be Plan). Id. at 290.
Part III(IV) of the injunction goes on to direct that Interior’s To-Be Plan identify any portions of the plan that might be deemed inconsistent with the common law trust duties previously identified by the district court, and explain why the identified portion or portions should not be considered inconsistent with these duties. Id. at 291.
Additionally, the court’s injunction required Interior to file with the Court, within 120 days, a “list of tribal laws and ordinances that the Interior defendants deem applicable to the administration of the Trust,” including “a full statement of the manner in which the Interior defendants consider these laws and ordinances to affect such administration.” Id. The court also ordered Interior to file within 90 days a detailed plan of measures it will take to correct certain “problems with the leasing, title, and accounting systems of the Trust,” and a plan identifying how Interior will “distinguish principal from income during [its] historical accounting of the Trust.” Id.
In Part IV(V) the court set forth a detailed timetable for implementing its order. The timetable not only covers requirements set forth elsewhere in the injunction, but also imposes several additional requirements on Interior, including several steps outlined in Interior’s Fiduciary Obligations Compliance Plan of January 6, 2003. Id. at 292-93. (The Compliance Plan was an early version of Interior’s plan to fulfill its fiduciary obligations and was subsequently replaced by the Comprehensive Plan. See id. at 243-44.) The court ordered that all of these requirements be completed within roughly three to six months. Id. at 292-93.
In Part V(VI) the court appointed a Judicial Monitor, endowed with “all authority bestowed on special masters pursuant to Rule 53” of the Federal Rules of Civil Procedure, “to report on the Interior defendants’ compliance with the provisions of this Order.” Id. at 294. According to the court, the monitor must have “unlimited access to the Interior defendants’ facilities and to all information relevant to the implementation of this Order.” Id. Finally, in Part VI(VII) the district court retained jurisdiction over the case until December 31, 2009. Id. at 295.
The government offers a number of reasons why we should vacate these provisions in their entirety (even to the extent that they are completely separate from “historical accounting”), as well as targeted arguments for vacating individual elements. We first reject two government arguments that, if sound, would call for vacating all “Fixing the System” aspects of the injunction. We then address the government’s argument that those elements violate the Supreme Court’s holdings in Lujan v. National Wildlife Federation, 497 U.S. 871, 110 S.Ct. 3177, 111 L.Ed.2d 695 (1990), and Norton v. Southern Utah Wilderness Alliance, — U.S. -, 124 S.Ct. 2373, 159 L.Ed.2d 137 (2004), which read the Administrative Procedure Act as limiting APA review to attacks on specific “agency action[s]” (or the unlawful withholding of such an action), and precluding its use for claims of broad programmatic failure. In light of this last argument, we reverse and remand for further action consistent with this opinion.
Government contentions applying to all elements of the injundion apart from historical accounting. Against the “Fixing the System” elements of the injunction, *100the government argues that (1) any consideration of trust deficiencies outside the realm of historical accounting represents an improper expansion of the lawsuit; and (2) under Mitchell I the government is not subject to any trust duties other than the statutorily created duty to account. We reject both contentions.
1. Expansion of the laivsuit. Interior claims that the district court cannot “expandf] its jurisdiction to include the entire field of trust management” because our decision in Cobell VI held “that the only actionable duty was the duty to perform an accounting.” Defendants’ Brief at 77. We made no such ruling.
First, we are puzzled by the idea that the “fixing” issues represent an expansion of the lawsuit. The complaint’s prayer for relief asked for an order “construing the trust obligations of defendants to the members of the class, declaring that defendants have breached, and are in continuing breach of, them trust obligations to such class members, and directing the institution of accounting and other practices in conformity [with the defendants’ trust] obligations.” Complaint at 26. It also claimed a wide range of past trust violations independent of accounting failures, e.g., that the government “[flailed to exercise prudence and observe the requirements of law with respect to investment and deposit of IIM funds, and to maximize the return on investments within the constraints of law and prudence,” and “[e]n-gag[ed] in self-dealing and benefiting from the management of the trust funds.” Complaint at 10. And at an early stage the district court responded to this range of attacks by bifurcating the case into the parts now before us — “fixing the system” and “correcting the accounts.” Scheduling Order at 2 (May 4,1998).
Interior misconstrues Cobell VI in arguing that our holding there limited the issue in this case to the provision of a historical accounting. We held that the duties identified by the district court, such as the duty to create specific written policies and procedures pursuant to the 1994 Act, 25 U.S.C. 162a(d)(6), were “subsidiary” to the duty to account, Cobell VI, 240 F.3d at 1105, not that the duty to account was the only fiduciary obligation in this case. “The 1994 Act did not create those obligations any more than it created the IIM accounts.... [The Act] ... recognized and reaffirmed what should be beyond dispute — that the government has longstanding and substantial trust obligations to Indians, particularly to IIM trust beneficiaries, not the least of which is a duty to account.” Id. at 1098 (emphasis added).
2. Statutory basis for fiduciary obligations. The government quotes United States v. Navajo Nation, 537 U.S. 488, 123 S.Ct. 1079, 155 L.Ed.2d 60 (2003), for the proposition that a purported trust beneficiary must “identify a substantive source of law that establishes specific fiduciary or other duties.” Id. at 506, 123 S.Ct. at 1091. The difficulty facing the government, however, is that, for the IIM accounts, such a duty is not far to seek.
In two matched pairs of cases the Supreme Court has stated what is needed to infer creation of conventional fiduciary duties with respect to Indian interests, sufficient to sustain claims for monetary damages under the Indian Tucker Act, 28 U.S.C. § 1505. (The modifier “conventional” is critical, to distinguish such duties from the concept that a trust relationship between the government and the Indians requires that statutory ambiguities be resolved in favor of Indians. See, e.g., Montana v. Blackfeet Tribe of Indians, 471 U.S. 759, 766, 105 S.Ct. 2399, 2403-04, 85 L.Ed.2d 753 (1985).) We described at the outset how in Mitchell I the Court found no enforceable fiduciary duty in the “trust” *101established for allotment lands themselves, given the limited purposes of the authority-retained by the government. Conversely, in United States v. Mitchell, 463 U.S. 206, 103 S.Ct. 2961, 77 L.Ed.2d 580 (1983) (Mitchell II), the Court found that where allotment land was subject to “elaborate [government] control” over property belonging to Indians, “a fiduciary relationship necessarily arises.” Id. at 225, 103 S.Ct. at 2972. Instead of the “bare” trust arising from the operation of the Dawes Act alone, id. at 224, 103 S.Ct. at 2971-72, the land involved in Mitchell II was subject to statutes and regulations asserting government control and responsibility, and compelling the inference of a genuine trust over the resources so controlled. A similar pair of cases applies the same principle to non-allotment land: see Navajo Nation, 537 U.S. at 507, 123 S.Ct. at 1091-92 (rejecting inference of enforceable fiduciary x-elationship because the statutes and regulations failed to give the government full responsibility to manage the resources in question for the bexxefit of the Indians), and United States v. White Mountain Apache Tribe, 537 U.S. 465, 123 S.Ct. 1126, 155 L.Ed.2d 40 (2003) (finding such a responsibility in the government).
The IIM accounts fall emphatically on the “full responsibility” side. Section 161a(b) directs that “[a]ll funds held in trust by the United States and carried in principal accounts on the books of the United States Treasury to the credit of individual Indians shall be invested by the Secretary of the Treasury, at the request of the Secretary of the Interior, in public debt securities with maturities suitable to the needs of the fund .... ” 25 U.S.C. § 161a(b). The statutory mandate, added in the 1994 Act, appears in large part to codify Interior’s prior practice, which involved the exercise of complete control over the IIM funds. See H.R.Rep. No. 103-778, at 11-12 (1994). Thus the statute assumes a set of funds “held” by the United States and directs its officials’ investment of these funds.
Another provision, 25 U.S.C. § 162a(a), authorizes an alternative investment for funds held in trust for the benefit of individual Indians — namely, deposits in banks selected by the Secretary of the Interior. And at the request of an individual Indian for whom funds are held, investments may also be made in obligations unconditionally guaranteed by the United States, or in mutual funds holding only such obligations. 25 U.S.C. § 162a(c). Although this extremely narrow band of permissible investments takes off the table many potential disputes over px'udent investment, it plainly assigns the government full managerial responsibility.
Under the four cases just discussed, these statutory mandates compel an inference of enforceable fiduciary duties. Indeed, the district court so held early in this litigation, see Cobell v. Babbitt, 52 F.Supp.2d 11, 22 (D.D.C.1999) (“Cobell III”) («q'hg basic contours of defendants’ fiduciary duties under this trust are established by the statutes [applicable to the IIM trust] and, as in Mitchell II, construed in light of the common law of tx-usts.”). Thus the trust duties that in Cobell VI we said the 1994 Act reaffirmed, 240 F.3d at 1100, see also id. at 1098, are the fully enforceable variety found in Mitchell II and White Mountain Apache Tribe.
That does not mean, however, that the district court may simply copy a list of common law trust duties from the Restatement and then order Interior to explain how it will satisfy them. Putting aside the litigation innovation (requiring defendants to explain how they will cure a long list of defaults as to which the court has made no evidence-based finding), the court has abstracted the common law duties from any statutory basis. Though the distx-ict court *102cites White Mountain Apache Tribe to support this incorporation of common law trust duties, see Cobell X, 283 F.Supp.2d at 265-67, it ignores the Supreme Court’s actual approach, which was to look to trust law to find that a particular common law duty — “to preserve and maintain trust assets” — was implied in a 1960 statute that, by permitting government occupation, made property “expressly subject to a trust,” White Mountain Apache Tribe, 537 U.S. at 475, 123 S.Ct. at 1133-34. Thus, once a statutory obligation is identified, the court may look to common law trust principles to particularize that obligation.
The district court itself so held in Cobell V, 91 F.Supp.2d at 38, finding that it could not grant plaintiffs’ prayer for a declaration of all trust duties arising from the IIM trust solely on the basis of plaintiffs’ common law trust claims. The court subsequently reversed itself on the point, saying that our decision in Cobell VI “supercedes” the district court’s prior observation that plaintiffs were wrong to think that once a trust relationship was established they could automatically “invoke all the rights that a common law trust entails.” Cobell X, 283 F.Supp.2d at 260 n. 12. Insofar as plaintiffs may have said that, they were wrong. In Cobell VI we actually held that the government’s duties must be “rooted in and outlined by the relevant statutes and treaties,” 240 F.3d at 1099, although those obligations may then be “defined in traditional equitable terms,” id.
Programmatic review under the APA. Plaintiffs invoke the APA as the basis for securing review of defendants’ conduct. Complaint at 26 (“Plaintiffs are entitled to review [of defendants’ various breaches of trust] under 5 U.S.C. § 702.”). Defendants argue that the district court’s “fixing the system” orders exceed the court’s jurisdiction because they are insufficiently pinned to discrete agency action (or inaction).
As Southern Utah notes, §§ 702, 704 and 706 of the APA “all insist upon an ‘agency action.’ ” 124 S.Ct. at 2378. This of course includes § 706(l)’s provision of authority to “compel agency action ... unreasonably delayed.” See id. at 2379 n. 1. Because of the requirement of specific agency action, the Court held initially in Lujan and again in Southern Utah that APA review was not available — even in the face of allegations of “rampant” violations of law — for claims seeking “wholesale improvement of [a] program by court decree, rather than in the offices of the Department or the halls of Congress, where programmatic improvements are normally made.” Lujan, 497 U.S. at 891, 110 S.Ct. at 3190; see also Southern Utah, 124 S.Ct. at 2380. The APA’s requirement of “discrete agency action,” Southern Utah explained, was
to protect agencies from undue judicial interference with their lawful discretion, and to avoid judicial entanglement in abstract policy disagreements which courts lack both expertise and information to resolve. If courts were empowered to enter general orders compelling compliance with broad statutory mandates, they would necessarily be empowered, as well, to determine whether compliance was achieved — which would mean that it would ultimately become the task of the supervising court, rather than the agency, to work out compliance with the broad statutory mandate, injecting the judge into day-to-day agency management .... The prospect of pervasive oversight by federal courts over the manner and pace of agency compliance with such [broad] congressional directives is not contemplated by the APA.
Id. at 2381.
The district court itself, earlier in this litigation, acknowledged the risk of taking *103on what were really legislative or executive functions: “The court has no present intention to entertain a request to sit as a pseuclo-congressional oversight body that tells defendants everything that they must do to meet their obligations programmatically. That is a role that only Congress can fulfill.” Cobell III, 52 F.Supp.2d at 31.
The application of Lujan and Southern Utah is complicated here by the availability of common law trust precepts to flesh out the statutory mandates, and, indeed, as we said in Cobell VI, at least partially to limit the deference that we would normally owe the defendants as interpreters of the statutes they are charged with administering. See Cobell VI, 240 F.3d at 1101. See also id. at 1104 (noting defendants’ obligation to “pass scrutiny under the more stringent standards demanded of a fiduciary”) (internal citation and quotation marks omitted).
The government accepts and even endorses our observation that interpretation of statutory terms is informed by common law trust principles, see Defendants’ Reply Brief at 26-27 (citing Cobell VI, 240 F.3d at 1099), but makes two key points as to why those precepts do not eliminate the risks that Lujan and Southern Utah saw in broad programmatic remedies. First, it notes that while the expenditures that plaintiffs seek are to be made out of appropriated funds, trust expenses for private trusts are normally met out of the trust funds themselves. Defendants’ Reply Brief at 27. Thus plaintiffs here are free of private beneficiaries’ incentive not to urge judicial compulsion of wasteful expenditures. Second, private trustees, even though held to high fiduciary standards, are generally free of direct judicial control over their methods of implementing these duties, and trustee choices of methods are reviewable only “to prevent an abuse by the trustee of his discretion.” Id. at 28 (citing Restatement (Second) of Trusts §§ 186-87 (1959)).
While a court might certainly act to prevent or remedy a trustee’s wrongful intermingling of trust accounts, this does not imply that the normal remedy would be an order specifying how the trustee should program its computers to avoid intermingling, as opposed to, for example, barring the use of a program that had caused forbidden intermingling or was clearly likely to do so. See Bogert & Bogert, Law of Trusts and Trustees § 861, p. 22 (“If the trustee has been given discretion with respect to the act in question, ... the court will not interfere by ordering him to take a certain line of conduct unless there is proof of an abuse of the discretion .... ”). “[A] court of equity will not interfere to control [trustees] in the exercise of a discretion vested in them by the instrument under which they act.” Firestone Tire and Rubber Co. v. Bruch, 489 U.S. 101, 111, 109 S.Ct. 948, 954-55, 103 L.Ed.2d 80 (1989) (internal quotation marks and citation omitted). The availability of the common law of trusts cannot fully neutralize the limits placed by the APA and the Court’s Lujan and Southern Utah decisions. Compare Cobell VI, 240 F.3d at 1104 (approving district court’s expression of intent to leave issue of choice of accounting methods, including statistical sampling, to administrative agencies), with Cobell X, 283 F.Supp.2d at 289 (forbidding use of statistical sampling).
That said, the question remains what specific elements of the “Fixing the System” decree run afoul of those decisions or are otherwise ill-founded. For the reasons explained below, we uphold the requirement to submit a plan and otherwise vacate and remand the case for further proceedings.
Plan. The core of Part III(IV) of the district court’s injunction is its order *104that Interior complete a detailed plan to fulfill its fiduciaiy obligations — specifically to fill in the as-yet inchoate To-Be Plan promised in the Comprehensive Plan. This command rests on the court’s prior order to file a Comprehensive Plan (issued in Cobell v. Norton, 226 F.Supp.2d 1, 162 (D.D.C.2002) (“Cobell VII”)), and on the district court’s finding here that the incompleteness of the To-Be Plan rendered the Comprehensive Plan only an interim step, see Cobell X, 283 F.Supp.2d at 284. The order thus in some respects continues or logically extends the original order to file the Comprehensive Plan. In Cobell VIII we upheld that order as a device to gather information for the court, “akin to an order ... relating] only to the conduct or progress of litigation.” 334 F.3d at 1138 (internal citation and quotation marks omitted). Thus, standing alone, the order to file the To-Be Plan simply enforces the prior order, which in effect required discovery of Interior’s plans consistent with the district court’s broad case management authority. To that extent we uphold it.
But Part III(IV) frames the plan by reference to the Interior defendants’ bringing themselves “into compliance with the fiduciary duties imposed upon trustees at common law, as identified by this Court in its memorandum opinions issued this date,” Cobell X, 283 F.Supp.2d at 291 (referring to sixteen specific common law trust duties enumerated by the court, id. at 267-71). And it requires Interior to “identify any portion of the To-Be Plan that might be deemed to be inconsistent with any of these fiduciary duties, and include a full explanation of why the identified portion or portions should not [be] considered to be inconsistent with any of these fiduciary duties.” Id. at 291. Finally, the district court ordered Interior to implement its plan. Id. at 290.
Thus the court evidently proposes to use the “plan” as a device for indefinitely extended all-purpose supervision of the defendants’ compliance with the sixteen general fiduciary duties listed. There are three difficulties with this approach.
First, the sole findings of unlawful behavior (other than accounting defaults) are stipulations acknowledging specific failures measurable against specific statutory mandates. See Cobell V, 91 F.Supp.2d at 32-34. See also Cobell VII, 226 F.Supp.2d at 66 (relying on the stipulations). The various plan filings can serve as the jumping-off point for judicial monitoring of Interior only to the extent that the monitoring is anchored either in these specific stipulations or in some future adjudicated findings. While in Cobell VI we upheld a requirement that the government produce periodic reports, we relied on specific findings by the district court “that appellants had unreasonably delayed the discharge of the[ir] duties by failing to ensure the provision of a complete historical accounting.” Cobell VI, 240 F.3d at 1107; see also Cobell V, 91 F.Supp.2d at 40 (finding commission of four specific accounting-related breaches of the 1994 Act). The district court cannot issue enforcement remedies— by any means — for trust breaches that it has not found to have occurred. The sixteen common law trust duties are pertinent only to the extent that they illuminate breaches already found (i.e., those named in the stipulations) or adjudicated in the future.
Second, the court’s innovation of requiring defendants to file a plan and then to say what “might” be wrong with it turns the litigation process on its head. However broad the government’s failures as trustee, which go back over many decades and many administrations, we can see no basis for reversing the usual roles in litigation and assigning to defendants a task that is normally the plaintiffs’ — to identify flaws in the defendants’ filings.
*105Third, in the absence of specific findings of unreasonable delay in Interior’s performance of its fiduciary duties, the court’s order that the defendants implement the entire Comprehensive Plan, including the full To-Be Plan, amounts to an order to obey the law in managing the trusts. Under this implementation order defendants would be subject to contempt charges for every legal failing, rather than simply to the civil remedies provided in the APA. See, e.g., NLRB v. Express Pub. Co., 312 U.S. 426, 435-36, 61 S.Ct. 693, 699-700, 85 L.Ed. 930 (1941) (“[T]he mere fact that a court has found that a defendant has committed an act in violation of a statute does not justify an injunction broadly to obey the statute and thus subject the defendant to contempt proceedings if he shall at any time in the future commit some new violation unlike and unrelated to that with which he was originally charged.”).
Finally, we note that the district court used language suggesting an intent to take complete charge of the details of whatever plan Interior might submit: “If the court [concludes that the plan will not satisfy defendants’ legal obligation], it may decide to modify the institutional defendant’s plan, adopt a plan submitted by another entity, or formulate a plan of its own that will satisfy the defendant’s liability.” Co-bell X, 283 F.Supp.2d at 142. This is in sharp contrast with Southern Utah’s point that “§ 706(1) empowers a court only to compel an agency ... to take action upon a matter, without directing hoiv it shall act.” 124 S.Ct. at 2379 (internal citation and quotation marks omitted).
In sum, while we uphold the district court’s order that Interior complete the To-Be Plan, we vacate the injunction insofar as it directs Interior, rather than the plaintiffs, to identify defects in its proposal and requires the agency to comply with the Comprehensive Plan.
Tribal laws and ordinances. The district court issued two directions about the trusts’ relations to such laws. In its “General Provisions,” it ordered the Interior defendants to “administer the Trust in compliance with applicable tribal law and ordinances.” Cobell X, Part H.D., 283 F.Supp.2d at 287. In a later section, it ordered them to compile a list of tribal laws and ordinances that they deemed applicable, with “a full statement of the manner in which the Interior defendants consider these laws and ordinances to affect such administration.” Part III(IV).C., id. at 291.
The first of these edicts — to apply tribal law to the extent applicable — appears meaningless, except as a general mandate to obey the law. It gains meaning, of course, because it is embodied in an injunction. Thus any violation is punishable by contempt, and the mandate is impermissible on the grounds stated above.
The instruction to list tribal laws deemed applicable poses a different issue. On its face it seems a specification not of Interior’s trust duties but of the court’s preferred methodology for assuring Interi- or’s fulfillment of those duties. As such it collides with the APA, Lujan, and Southern Utah. It may be helpful for defendants in fulfillment of their trust duties to compile such a list (perhaps including tribal provisions on title, ownership, leasing, and contract for the purposes identified by the district court, Cobell X, 283 F.Supp.2d at 275, or provisions on inheritance, see Felix S. Cohen, Handbook of Federal Indian Law 634 (1982 ed.)). But a list of applicable tribal laws is no more essential to ensure that Interior “accelerated” rather than “delay[s]” fulfillment of its fiduciary obligations, see Cobell X, 283 F.Supp.2d at 275, than would be a list of all federal and state laws with which Interior must comply in administration of the IIM trusts. *106Although the district court may declare the government’s legal obligations whether rooted in federal or tribal law — pursuant to the Declaratory Judgment Act, 28 U.S.C. § 2201, see Cobell V, 91 F.Supp.2d at 38, it may not prescribe the specific steps the government must take to comply with these obligations unless it has found that government actions (or inactions) breached a legal duty and that the steps ordered by the court constituted an essential remedy.
Appointment of a court monitor. In Part V(VI), the court “appointed] a Judicial Monitor to report on the Interior defendants’ compliance with the provisions of this Order.” Cobell X, 283 F.Supp.2d at 294. “The Judicial Monitor shall be appointed pursuant to Rule 53 of the Federal Rules of Civil Procedure, and shall possess all authority bestowed on special masters pursuant to Rule 53.” Id. (emphasis added). Thus the label “Monitor” is inaccurate; the authority purportedly bestowed is really that of a “Master.” Whereas a monitor’s “primary function is to monitor compliance,” a master’s role is broader: to “report[ ] to the court and, if required, make[ ] findings of facts and conclusions of law.” Special Project, The Remedial Process in Institutional Reform Litigation, 78 Colum. L. Rev. 784, 827-28 (1978) [hereinafter “Special Project”]-, see also id. at 829 (“[A] monitor’s activities are so unlike those of a rule 53 master that the court should not [designate a monitor a master] .... Monitoring rarely, if ever, proceeds by the quasi-judicial hearings envisaged by rule 53.”). The district court also specified that “Interior defendants shall provide the Judicial Monitor and his or her agents with unlimited access to the Interi- or defendants’ facilities and to all information relevant to the implementation of this Order, in order that the Judicial Monitor and his or her agents may be made cognizant of any failures to comply with the provisions of this Order.” Cobell X, 283 F.Supp.2d at 294.
According to the Interior defendants, the appointment of a monitor exceeds the scope of the district court’s authority. We agree.
In April 2001 the government consented to the appointment of a court monitor for one year. In April 2002, notwithstanding the government’s objection, the district court reappointed the court monitor, a decision we reversed in Cobell VIII. In rejecting the monitor, we wrote: “The Monitor’s portfolio was truly extraordinary; instead of resolving disputes brought to him by the parties, he became something like a party himself. The Monitor was charged with an investigative, quasi-inquisitorial, quasi-prosecutorial role that is unknown to our adversarial legal system.” Cobell VIII, 334 F.3d at 1142. We distinguished the monitor in this case from the permissible appointment of a master in Ruiz v. Estelle, 679 F.2d 1115, 1161-62 (5th Cir.) (prison reform), amended in part, reh’g denied in part on other grounds, 688 F.2d 266 (5th Cir.1982). We explained:
The role of the special master in Ruiz was not nearly as broad as the role of the Monitor in this case. There the master was specifically instructed “not to intervene in the administrative management of [the department] and ... not to direct the defendants or any of their subordinates to take or to refrain from taking any specific action to achieve compliance.” [679 F.2d] at 1162. Most important, the court of appeals clarified that the special master and the monitors were “not to consider matters that go beyond superintending compliance roith the district court’s decree,” thereby assuring the special master would not be an “advocate” for the *107plaintiffs or a “roving federal district court.” Id.
334 F.3d at 1143 (emphasis added).
Unlike the monitor in Ruiz, we said, the monitor appointed in 2002 could not “have been limited to enforcing a decree, for there was no decree to enforce, let alone the sort of specific and detailed decree issued in Ruiz and typical of such cases.” Id.
In appointing a monitor in Cobell X, the district court adopted almost verbatim the language we used to explain that the court monitor in Ruiz was permissible because of its circumscribed role. According to the district court:
The Judicial Monitor and his or her agents shall not intervene in the administrative management of the Interior defendants. The Judicial Monitor and his or her agents shall not direct the Interior defendants or any of their subordinates to take or to refrain from taking any specific action to achieve compliance with this Order. The Judicial Monitor and his or her agents shall not consider matters that go beyond superintending or reporting upon compliance tvith this Order.
283 F.Supp.2d at 295 (emphasis added).
Despite the similarity of the language we used to distinguish Ruiz and the language used by the district court to limit the monitor’s authority, there is a significant difference between the two cases. The “Fixing the System” part of the present injunction (especially given the excisions already discussed) is not nearly as complex as the specific relief ordered in Ruiz (embodied partly in two consent decrees appearing at Ruiz, 679 F.2d at 1127-28, 1165-68, 1174-84, partly in a hotly contested order summarized id. at 1164). If at some future time the non-accounting aspects of the case culminate in a true remedial injunction with specific duties tied to specific legal violations cognizable under the APA, the usual latitude for masters to oversee compliance would come into play. See United States v. Microsoft, 147 F.3d 935, 954 (D.C.Cir.1998). Alternatively, appointment of a true judicial monitor, with duties focused on determining just how defendants’ management of their trust duties is proceeding, might become appropriate. “Monitors are appropriate if the remedy is complex, if compliance is difficult to measure, or if observation of the defendant’s conduct is restricted.” Special Project, 78 Colum. L. Rev. at 828. Compare Cobell X, 283 F.Supp.2d at 218 (observing that Interior’s quarterly reports have given an overly optimistic and inaccurate portrait of their reform efforts).
Additional provisions. The injunction imposes several additional duties on defendants. For example, the court revived elements of Interior’s Compliance Plan, which was replaced by its Comprehensive Plan, Cobell X, 283 F.Supp.2d at 244, to require Interior to “request legislation from Congress to satisfy part of its imbalance of Trust fund balances with” Treasury. Id. at 292. The court also ordered Interior to “request an expansion of the fiscal year 2004 annual audit to include all funds held in trust by the United States for the benefit of an individual Indian” and invested pursuant to 25 U.S.C. § 162a, id., among numerous other requirements. Thus, rather than acting to assure that “agency action” conforms to law, the court has sought to make the law conform to the court’s views as to how the trusts may best be run. The limits on the court’s remedial authority, discussed at length above, apply equally to these additional requirements in the injunction. The court’s authority is limited to considering specific claims that Interior breached particular statutory trust duties, understood in light of the common law of trusts, and to ordering specific relief for those breaches. To the *108extent Interior’s malfeasance is demonstrated to be prolonged and ongoing, more intrusive relief may be appropriate, as we held was the case in Cobell VI for the government’s failure to provide a statutorily required accounting. Yet the court may not micromanage court-ordered reform efforts undertaken to comply with general trust duties enumerated by the court, and then subject defendants to findings of contempt for failure to implement such reforms.
:¡: ‡ * . * * *
The “historical accounting” elements of the injunction are vacated because of the mandate of Pub.L. No. 108-108, and the remainder of the injunction, aside from the requirement that Interior complete its To-Be Plan, is vacated and remanded to the district court for revisions not inconsistent with this opinion.
So ordered.