OJSTI Case Report 2026-02: forensic scope, analytic methodology, and evidence basis
This analytic brief documents the forensic scope, analytic methodology, and evidence-handling model used by OJS One Threat Intelligence to assess a compromised OJS client environment. It derives from the same anonymized engagement covered in the primary case report, but focuses on method: how evidence was structured, validated, and bounded.
What this report covers
- How evidence was organized into three independent source groups.
- Which validation criteria were used to support confirmed findings.
- How filesystem, HTTP/FTP logs, and SQL material were reviewed in a traceable workflow.
- Which conclusions were classified as confirmed, probable, or undetermined.
- Why source separation is critical in multi-journal OJS environments.
Purpose of this report
This report exists to make the analytic basis of the case explicit. In high-credibility threat intelligence, it is not enough to state that malware or suspicious activity was observed. The report must also show how conclusions were reached, what evidence classes were involved, and where confidence boundaries remain.
In OJS ecosystems, this is especially important because environments often include multiple journals, many SQL dumps, overlapping paths, and heterogeneous artifacts collected from backups, shared hosting, or fast incident-response actions.
Corpus definition and source separation
The case was analyzed through three evidence groupings.
1. Client hosting evidence set
The primary hosting corpus included a broad account-level dump with web content under public_html, sensitive account directories under homedir, compressed HTTP and FTP logs, and multiple MySQL dumps across OJS instances.
This corpus was the main source for:
- malicious PHP discovery in active OJS paths;
- timestamp review;
- HTTP validation of shell usage;
- FTP write-event review;
- broad SQL screening for suspicious accounts and content patterns.
2. Client backup evidence set
The second corpus was a curated backup-derived package containing extracted malicious files, SHA-256 hashes, IOC summaries, numbered snippets, SQL extracts related to suspicious users and profile contamination, and a dedicated suspicious-account report.
It was especially valuable for:
- validating exact malicious code behavior in selected files;
- preserving technical hash traceability;
- documenting backup-derived database indicators;
- enabling reproducible artifact citation.
3. Auxiliary forensic note set
A smaller third corpus existed as an auxiliary forensic note set. Based on available material, its direct technical contribution was limited. It was therefore treated as contextual support, not as a primary evidentiary basis for core conclusions.
This source-separation model matters because evidence sets do not carry equal weight. Live-path artifacts and original logs are direct observations. Backup packages can strongly support those observations, but they are not equivalent to full live acquisition.
Evidence handling and validation criteria
The review followed a strict but practical confidence hierarchy.
Directly observed evidence
This category included:
- PHP files reviewed directly in the hosting corpus;
- records extracted from compressed HTTP and FTP logs;
- file metadata and timestamps directly visible in available material;
- SQL rows or patterns directly observed in dump files.
These findings were treated as the strongest basis for confirmed statements.
Curated technical evidence
This category included:
- extracted malicious files in the backup package;
- hash inventories;
- numbered snippets;
- curated SQL extracts and IOC lists.
These materials were treated as valid technical evidence when content matched what could be independently verified, or when internal consistency was high and no conflicting primary evidence existed.
Secondary summary evidence
This category included:
- summary markdown notes;
- pre-written incident summaries;
- auxiliary forensic notes.
These materials were useful for orientation and scoping, but did not qualify as confirmed findings unless supported by direct artifact review, logs, hashes, or strong cross-corpus corroboration.
Filesystem analysis methodology
Filesystem review focused on identifying PHP components inconsistent with standard OJS operation and determining whether those components occupied plausible application paths.
The approach included:
- path-based review of OJS directories and adjacent web content;
- pattern matching for webshell indicators, remote loader logic, embedded authentication, suspicious upload routines, and cloaking behavior;
- cross-path comparison to identify duplicated tooling;
- timestamp review to identify suspicious temporal clustering;
- path normalization to report findings in standardized OJS-relative format.
This methodology identified several malicious artifact classes:
- remote loaders using external payload retrieval and
eval(); phpFileManagercopies with hardcoded credentials;- upload-capable PHP shells;
- a bootstrap-level cloaking component targeting Google-related user agents.
Log analysis methodology
Log review determined whether malicious server-side artifacts were only present or operationally used.
The workflow included:
- extraction and review of compressed HTTP logs for exact requests to suspicious PHP paths;
- correlation of request timing with malicious filenames and file timestamps;
- review of status codes and request sequences to identify successful interactive use;
- extraction and review of FTP logs for write activity in sensitive OJS locations;
- clustering of repeated source-IP behavior where analytically relevant.
The key contribution of this stage was moving from static artifact discovery to confirmed operational usage, including file-manager tree enumeration of /home/ via shell paths relative to OJS.
Database review methodology
Database review was necessary because filesystem compromise alone did not explain all observed indicators.
The process included:
- screening SQL dumps for suspicious usernames, emails, domains, profile text, signatures, biographies, and spam-related strings;
- review of temporal concentration in account creation activity;
- targeted review of rows highlighted in backup-derived material;
- distinction between clearly suspicious patterns and low-confidence anomalies;
- explicit avoidance of overclaiming where generic contact data could have benign explanations.
This supported cautious but useful conclusions:
- some account clusters were highly suspicious;
- some profile content showed clear contamination patterns;
- not every anomalous record should be classified as malicious without criteria-based justification.
Analytic standards and confidence model
The case used a standards-based distinction between facts, assessments, and unresolved questions.
Confirmed
Statements were treated as confirmed when supported by one or more of:
- direct code review;
- verifiable log entries;
- matching artifacts across corpora;
- consistent hash and snippet evidence;
- timestamp or SQL evidence directly tied to observed artifact sets.
Probable or partial
Statements were treated as probable when evidence strongly suggested unauthorized or malicious activity but did not fully establish mechanism, actor behavior, or total scope. Suspicious account creation and part of database abuse remained in this category unless tightly constrained and well documented.
Not determined
Statements remained outside confirmed assessment where available material did not allow defensible closure, including:
- definitive initial access vector;
- actor identity or attribution;
- confirmed exfiltration;
- full blast radius across every OJS instance in the account;
- whether all suspicious database records belonged to the same intrusion wave.
This confidence model prevented claim inflation while preserving technical value.
MITRE ATT&CK assessment
In this methodology report, ATT&CK is used as an evidence discipline mechanism, not as a substitute for proof.
Methodologically relevant mappings:
T1505.003 - Server Software Component: Web Shell
Applied where server-side PHP tooling clearly provided unauthorized operational capability.T1059.006 - Command and Scripting Interpreter: PHP
Applied where malicious logic executed through PHP scripts under web-accessible paths.T1105 - Ingress Tool Transfer
Applied where upload logic or remote payload retrieval was directly observed.T1036 - Masquerading
Applied where malicious files were placed in plausible OJS-relative paths or disguised through naming.T1136 - Create Account
Used cautiously and only where database evidence justified discussion of suspicious account presence or likely unauthorized creation.
Cyber Kill Chain assessment
The same discipline applies to Cyber Kill Chain. Here, the framework describes which phases the available evidence can meaningfully illuminate.
The corpus most clearly supports:
- Installation, through malicious server-side file placement;
- Command and Control / Operational Control, through interactive shell use and remote loader behavior;
- Actions on Objectives, through file manipulation, cloaking, and application-layer abuse indicators.
Earlier delivery and exploitation stages remained less visible because available evidence was stronger in post-access artifacts than in initial intrusion telemetry.
Evidence limitations
- The backup-derived package is technically valuable, but not equivalent to full live forensic acquisition.
- The hosting corpus spans multiple OJS instances, which increases coverage but complicates scoping.
- Some log review was necessarily selective due to volume.
- SQL review required pattern-based triage before deeper inspection.
- The auxiliary forensic note set contributed little direct technical value.
- Source anonymization is appropriate for publication, but it limits public exposure of all raw paths and domains observed in underlying client material.
These limitations do not invalidate the compromise finding, but they define confidence boundaries for specific claims.
Conclusion
The methodology was sufficient to support a high-confidence technical compromise assessment while preserving a defensible distinction between confirmed evidence, strong inference, and unresolved questions.
The key strength of the workflow was not a single artifact family, but disciplined combination: filesystem review, log correlation, database screening, and cross-corpus validation.
For OJS-focused threat intelligence, that distinction matters: high-quality reporting depends as much on evidence handling as on malware discovery.
OJS ONE applies this same methodology in its OJS cybersecurity services: defined scope, preserved evidence and conclusions stated with their confidence level. If you suspect a compromise, contact us before deleting files or reinstalling.
Appendix: source inventory
Primary evidence categories
- Hosting-side web content and OJS paths
- Compressed HTTP logs
- Compressed FTP logs
- MySQL dumps across multiple OJS instances
- Backup-derived malicious file set
- Backup-derived hash inventory
- Backup-derived IOC material
- Backup-derived SQL extracts
- Suspicious-account assessment
Standardized OJS-relative paths referenced in this case
bootstrap.inc.phpplugins/generic/pdfJsViewer/pdfJsViewer.phpapi/v1/users/UserHandlerr.phppages/manager/sans.phptools/uplaodsers.phptemplates/manageIssues/DebugFilter.phptools/Index.phptools/SearchIndex.phpapi/v1/submissions/core.phpcontrollers/modals/submissionMetadata/linkAction/sans.php
About OJS One
OJS One is a specialist firm focused on Open Journal Systems, providing support, infrastructure, operational continuity, and cybersecurity services for OJS-based publishing environments.
Its approach combines day-to-day platform operations with compromise assessment, hardening, incident response, and applied analysis of editorial-environment integrity.
Through OJSTI, OJS One develops technical documentation focused on compromise patterns, observed techniques, and defensive lessons relevant to academic institutions, editorial teams, and security stakeholders operating OJS.
Is your journal struggling with slow performance or recurrent outages?
Contact our team and see how enterprise-grade infrastructure can transform your editorial operation.
Request assessment