The internal Federal Police report used by Minister Alexandre de Moraes, of the STF (Supreme Federal Court), to counterattack his court colleague, Minister André Mendonça, contains information classified as low confidence and complaints against Jorge Messias, the Union's attorney general.
The document, with no identified author, analyzes Mendonça's behavior as rapporteur of inquiries related to Banco Master and INSS (National Social Security Institute) deviations, as well as his relationship with other authorities.
Based on the document, classified by the PF itself as "without probative value", Moraes accuses Mendonça of having focused the investigations against him and other authorities — such as the president of the Senate, Davi Alcolumbre (União Brasil-AP) —, while sparing others being investigated "for personal and political reasons".
See the main points of the document:
Low confidence information
Some of the most serious hypotheses in the report are classified as conjectures of low reliability. This is the case, for example, of the assertion that Mendonça's decisions were part of a "strategy of recomposition of forces" in the STF, that he acted in political collusion with Jorge Messias or that he aimed to facilitate an amnesty for those convicted of the coup attacks on January 8th.
The report itself directly warns that it "does not authorize institutional affirmation, formal action or external manifestation" and that the public use of these scenarios could "contaminate" the validity of other solid investigative acts.
Davi Alcolumbre as a strategic target
The report classifies senator Davi Alcolumbre as a "probable strategic target" of Mendonça. The hypothesis formulated attests that the minister had a "known disagreement" with the senator since the last government, when the parliamentarian acted as a "major obstacle" to his nomination to the STF.
Mendonça's plan would consist of using the national president of União Brasil, Antônio Rueda as an "access route" to reach and weaken Alcolumbre. According to the police thesis, the magistrate intended to destabilize him politically to prevent the maintenance of favoritism for the presidency of the Senate.
The ultimate objective of this maneuver would be to prevent Alcolumbre from controlling the voting agenda for possible requests for impeachment of STF ministers, making it possible, in theory, for a future recomposition of forces in the Supreme Court in favor of amnesty for those convicted of the January 8 coup attack.
Alliance with Jorge Messias
The Union's attorney general is cited in the context of the AGU's decision not to appeal against Mendonça's decisions that barred the PF's administrative acts — such as restrictions on the composition of investigative teams and the delivery of evidence.
The PF assessed that the AGU's decision not to take action against the reporting minister came from Messias' personal interest in "preserving the political relationship" with Mendonça. As his nomination to the STF was rejected by the Senate in April 2026, Messias would be seeking the sympathy and support of Mendonça and his bases of influence (such as evangelical churches, given the "common religious matrix" of both) to try to renew his nomination to the court.
The PF itself warns that the assessment had "LOW aggregate confidence" and did not authorize any formal action or external manifestation by the corporation, serving exclusively for "monitoring and targeted collection" of internal intelligence.
Different treatment to Moraes
According to the document, Mendonça demonstrated a personal interest in starting a formal investigation against Moraes. The magistrate would have asked, more than once, that the Federal Police team "send some provocation to him in this regard".
As neither the PF nor the PGR presented elements against Moraes, Mendonça acted ex officio. According to the material, the minister ordered extensive investigative action against his colleague and called the investigators to a meeting in his office, in which he delivered an unsigned copy of his decision without prior consultation with the Attorney General's Office, a procedure that would be outside the ordinary procedural channels.
The document states that there was a difference in treatment of Moraes in relation to other ministers, such as Gilmar Mendes, Luiz Fux, Dias Toffoli and Kassio Nunes Marques, which would be a relevant indicator.
"As for the cause, the hypothesis of correlation with affinities internal to the collegiate and with the origin of nomination for the position remains under observation, without support in current elements and without the need for a risk judgment: for management purposes, the observable asymmetry is important, not its motivation", says the text, also classified with a low level of confidence on this point, as there is no documentary comparison constructed.
Distrust about Gonet and Andrei Rodrigues
The report used by Moraes mentions by name the Attorney General of the Republic, Paulo Gonet, and the Director General of the PF, Andrei Rodrigues, only once. He deals with what he calls "generic imputations" directed at the two, given Mendonça's suspicions about both.
The text reports that Mendonça repeatedly stated, in meetings, that Andrei maintains "inadequate proximity" with President Lula (PT) and, therefore, "it would not be possible to trust him". There is also a report that, in a meeting with the investigation team, Mendonça declared that he had prepared a procedure in which he considered determining the precautionary removal of Andrei Rodrigues from his duties.
Regarding Gonet, the report points out that Mendonça applied the same suspicion, in this case, in reference to the relationship between the head of the agency and Gilmar Mendes. Here too, this would make him "untrustworthy" and, therefore, the minister would have considered calling Gonet as a witness in the proceedings. For the PF, this is a maneuver to remove him from command of the prosecution of the case.
The PF classified the information with a high degree of confidence regarding the absence of a concrete fact invoked by Mendonça and regarding the legal consequences of the measures considered, but classified it as low to moderate regarding the occurrence of pressure meetings, as they were based only on verbal reports.
Mendonça's interference in investigations
The PF report states that Mendonça, as rapporteur of the INSS and Master cases at the STF, assumed an investigative role in episodes such as choosing delegates to act on the case and accessing the evidence collection.
"It is assessed that the reporting court has been exercising, in Operation Without Discount and in related events, powers typical of presiding over the investigation, and not of judicial supervision, with three convergent effects: subordination of administrative management acts of the Federal Police to judicial control; direct access to the raw evidence collection; and suppression of the delegate's technical supervision over the unit's analytical product", states the report.
This is one of the points presented by the STF wing for Mendonça to abandon the case, which includes President Lula's son among those being investigated.
Difference in treatment between investigated
In another part of the document, the PF points out the time difference between the representation made by investigators in cases involving senators Jaques Wagner (PT-BA) and Ciro Nogueira (PP-PI).
While the analysis of search requests against Wagner took nine days, in Ciro's case it took 29 days. Against former governor Cláudio Castro (PL-RJ), the deadline was even longer, 75 days.
The operation against Wagner was launched in June and resulted in his removal from his position as government leader in the Senate. Ciro Nogueira was targeted in May. Mendonça and Nogueira were ministers of Jair Bolsonaro (PL).
"The dispersion is significant and suggests a correlation between assessment time and the profile of the person being investigated", states the report.
The text, however, emphasizes that the degree of confidence in the hypothesis is low, due to a selective sample of targets, with uncontrolled variables (such as complexity and days worked).
"The series, however, is small and uncontrolled: heterogeneous measurements in volume and complexity; intervals that include the time spent viewing the PGR, not attributable to the judgment; recesses and peaks in the agenda not isolated. The data, in its current state, is indicative."
