AURA: A Multi-Agent Intelligence Framework for Knowledge-Enhanced Cyber Threat Attribution

Nanda Rani; Sandeep Kumar Shukla

REVIEW 3 major objections 6 minor 39 references

AURA: A Multi-Agent Intelligence Framework for Knowledge-Enhanced Cyber Threat Attribution

T0 review · 3 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read AURA shows that a retrieval-augmented multi-agent framework can attribute APT campaigns to known threat groups, reaching 63.33% top-1 group-level accuracy and 100% top-2 nation-level accuracy on 30 held-out reports.

desk verdict The agent architecture is a real combination, but the evaluation lets the system retrieve the answer from related reports rather than reason toward it, so the accuracy claims are unproven. read the letter →

arxiv 2506.10175 v1 pith:SLQN7ASM submitted 2025-06-11 cs.CR

Nanda Rani , Sandeep Kumar Shukla This is my paper · ORCID

classification cs.CR

keywords ThreatAttributionAdvancedPersistentThreatsRetrieval-AugmentedGenerationMulti-AgentSystemsLargeLanguageModelsCyberIntelligenceTacticsTechniquesandProceduresInterpretability

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Attributing a cyber attack to the responsible state-sponsored group is usually a slow, manual job that mixes malware artifacts, infrastructure clues, and judgment calls. This paper claims that much of that reasoning can be delegated to a team of cooperating LLM agents that rewrite the analyst's query, retrieve relevant threat reports from a vector database, filter out irrelevant context, and then produce both an attributed actor and a plain-language justification. The authors evaluate the framework on 30 APT reports deliberately selected to postdate the LLMs' training cutoffs and report up to 63.33% top-1 and 73.33% top-2 group-level accuracy, with 100% top-2 accuracy at nation level for one model. If these numbers hold, AURA would give threat analysts a transparent triage tool that surfaces plausible actors and the evidence behind them, rather than a black-box label.

What carries the argument

The load-bearing mechanism is the agent pipeline itself, expressed as the transformation $\mathrm{AURA}(Q)=(A,J)$. Six components do the work: a preprocessing module that extracts TTPs, IoCs, malware names, and timelines; a semantic retriever that ranks chunks of the threat-report corpus by cosine similarity of dense embeddings; a decision agent that discards irrelevant retrieved context; a query-rewriting agent that resolves pronouns and vague references before retrieval; an attribution-generation agent that invokes the LLM over the rewritten query and retrieved evidence to pick the actor; and a conversational-memory module that carries context across turns. A web-search module exists but is disabled during evaluation to prevent leakage, so the reported numbers test only internal knowledge-base retrieval.

What would settle it

Run a membership test on the 30 test reports against each of the four models, or prompt the models to complete or summarize a test report with retrieval disabled; if any model reproduces report-specific details or attains comparable attribution accuracy without the retrieval context, the measured accuracy does not isolate the framework's contribution.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that a retrieval-augmented multi-agent pipeline can turn unstructured threat intelligence into an interpretable attribution decision: the framework maps a natural-language query $Q$ to a predicted threat actor $A$ and a justification $J$. The pipeline decomposes the task into query rewriting, semantic retrieval over a corpus of 2,199 threat reports, a relevance decision gate, attribution generation, and justification synthesis, with the attribution step carried out by one of four proprietary LLMs. Across 30 held-out test reports, the best configuration reaches 63.33% top-1 and 73.33% top-2 group-level accuracy, and one model reaches 83.33% top-1 and 100% top-2 nation-level accuracy, with justifications rated fluent and coherent by an LLM acting as judge. The authors take these results as evidence that agent modularity plus retrieval grounding improves both accuracy and explainability compared with single-pass LLM extraction or ML classifiers used in prior work.

Load-bearing premise

The results stand on the unverified assumption that none of the 30 test reports appeared in the training data of the four evaluated LLMs; the paper relies on publication dates versus knowledge cutoffs but never checks this per model.

Editorial extensions

If this is right

Nation-level attribution is consistently easier than group-level attribution in these experiments, so the framework is most reliable as a geopolitical triage filter rather than a definitive actor identifier.
Surfacing two candidates instead of one raises accuracy by roughly ten points at group level, which matters because real APT groups often share tooling and TTPs.
The framework is provider-agnostic: swapping the final attribution LLM changes accuracy but leaves the pipeline intact, making the design portable to newer or open-weight models.
With web search disabled, the reported numbers are a conservative floor; deployment with live retrieval is expected to perform at least as well on previously unseen intelligence.
Generated justifications give analysts a traceable path from evidence to actor, which is the property earlier ML-based attribution systems lacked.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A direct ablation the paper does not run — attribution with retrieval disabled — would separate the contribution of the RAG pipeline from the parametric knowledge of the LLM; without it, the 63% figure mixes both sources.
Because the test set has only 30 reports, the gap between the best model (63.33%, 19/30) and the second-best (53.33%, 16/30) is within a few correct answers of being noise; larger testbeds are needed before treating model rankings as stable.
The same agent architecture should transfer to intrusion-set-level or campaign-level attribution, since the retriever and justification modules do not depend on group-level labels.
One can test the value of the decision agent by feeding the attribution agent unfiltered top-k chunks; if accuracy does not drop, the relevance gate is not load-bearing.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Colleague,

The genuinely new thing here is the agent orchestration: query rewriting, relevance gating, conversational memory, and justification synthesis combined for APT attribution. I haven't seen that combination in prior work, and the system is described clearly enough to be reproduced. The authors are also honest that the test set is small and that they plan open-source evaluation later.

The problem is the evaluation, and it's load-bearing. The headline accuracy numbers—63.33% group top-1, 73.33% top-2, up to 100% nation top-2 on 30 reports—do not demonstrate that AURA is reasoning over TTPs and IoCs. The held-out split removes the 30 test reports from the vector DB, but the residual 2,199-document corpus is drawn from the same GitHub collections (APTNotes, CyberMonitor) and includes multiple vendor reports on the same campaigns. Retrieval can therefore return a chunk that explicitly names the ground-truth actor. The case study makes this concrete: AURA's justification cites a Cyfirma report that directly links the campaign to APT36. Disabling web search does not address this internal leakage. So the measured accuracy may be lookup, not inference.

There are also standard-but-real issues: no baseline comparison to prior attribution systems (Guru et al., NO-DOUBT, the authors' own TTP-based method), no confidence intervals on 30 samples, and an LLM-as-Judge using gpt-4o, same model family as the generator. The paper acknowledges the small test set, but not the internal leakage.

That said, the framework is plausible and the modular design is a reasonable step toward interpretable attribution, which is a real need. The fix is straightforward: build a test set of campaigns not represented in the knowledge base at all, compare against a simpler RAG baseline and a single-pass LLM, and get human expert ratings on justifications. With that, the contribution becomes credible.

I would send this to review because the idea deserves referee time, but I'd expect major revision. The stress-test is correct; raise it front and center.

— [Your name]

Referee Report

3 major / 6 minor

Summary. This paper introduces AURA, a multi-agent retrieval-augmented framework for attributing APT campaigns to threat groups and nation-states. The system preprocesses analyst queries, rewrites them, retrieves chunks from a 2,199-report vector database, uses a decision agent to filter irrelevant context, optionally performs web search (disabled in evaluation), and generates an attribution decision with a natural-language justification. The authors evaluate four proprietary LLMs on 30 held-out reports, reporting up to 63.33% top-1 and 73.33% top-2 group-level accuracy for GPT-4o, and up to 100% top-2 nation-level accuracy for Claude 3.5 Sonnet. Justification quality is assessed with readability, lexical richness, coherence, perplexity, and an LLM-as-judge protocol; two case studies are also presented. A limitation section acknowledges the small test set.

Significance. If the accuracy results survive a leakage-controlled re-evaluation, AURA would be a useful step toward transparent, interpretable APT attribution: the modular agent design, dual group/nation granularity, and explicit justifications are valuable for analyst-facing decision support, and using a held-out set of 30 recent reports is a reasonable start given the difficulty of obtaining post-training data. The decision to disable web search during evaluation and the paper's candid limitation statement are strengths. However, the contribution currently rests on a narrow empirical base: there are no quantitative baselines, no uncertainty quantification, and no human validation of the justifications. The framework itself is an engineering combination of existing RAG and LLM components, so its value depends on the empirical claims, which are not yet established.

major comments (3)

[§3.3–3.4, §4.1, §5, §6.2.1] The held-out split only removes the 30 test report files from the vector database, not the many other reports in the same GitHub collections (APTNotes, CyberMonitor) that describe the same campaigns or actors. The Semantic Retriever returns top-k chunks by cosine similarity, and the Decision Agent filters for generic relevance, not for whether a chunk already contains the ground-truth actor name. The Section 6.2.1 case study is direct evidence: the generated justification says "Public threat intelligence by Cyfirma [24] directly links this campaign to APT36," so the retrieved context includes a document with the label. Because both the training corpus and the test set are drawn from the same collections, other vendor write-ups for the same campaign are likely in the 2,199-document corpus. This makes the headline 63.33%/73.33% numbers potentially a lookup effect rather than evidence of reasoning over TTPs and IoCs. The authors should report, for each of the 30 test reports, whether any top-k chunk returned before the Decision Agent contains the ground-truth group or nation name, and should re-run the evaluation with a knowledge base pruned of all reports that mention those labels or describe the same campaign.
[§5, §4.1] The central accuracy claims rest on only 30 test reports, with no confidence intervals, significance tests, or baselines. With n=30, the difference between GPT-4o's 63.33% and Claude 3.5 Sonnet's 53.33% top-1 group accuracy is about three reports, so the reported ranking of models is not statistically supported. There is also no quantitative comparison: no no-retrieval LLM prompt, no single-agent or naive-RAG ablation, and no comparison with NO-DOUBT, DeepAPT, or the prior TTP-based methods cited in Table 1. The sentence in Section 6 that "the integration of semantic retrieval with task-specific knowledge effectively grounds the attribution process" is therefore not demonstrated by the presented experiments. In addition, the pretraining-cutoff argument in Section 4.1 is asserted but not verified for each of the four proprietary models; the paper's own limitation section acknowledges the small test size. A minimal fix is to provide per-report predictions with bootstrapped confidence intervals, a retrieval-free baseline, and a rerun with a pruned knowledge base.
[§6.1.2, Abstract] The claim of "expert-aligned justifications" is unsupported. The LLM-as-judge is GPT-4o, which is the same model family used for the attribution agent in most experiments and for all non-attribution agents in every experiment (per Section 3.9); this is a self-referential evaluation, not an independent expert assessment. The automated metrics in Section 6.1.1 (readability, lexical richness, embedding coherence, perplexity) measure textual surface properties, not attribution correctness or alignment with expert reasoning. The abstract and Section 6.1.2 use the phrases "expert-aligned" and "human-aligned evaluation," but no human annotators are involved. The authors should either add a human evaluation with multiple expert annotators, or reframe the claim as an LLM-as-judge quality check and remove "expert-aligned" from the abstract.

minor comments (6)

[§3.1] The text says the framework comprises "six key components" but then enumerates seven items (i)-(vii); please correct the count or the list.
[§4.3, §5] The setup section says AURA is evaluated using the pass@3 metric, but Section 5 reports only top-1 and top-2 accuracy; please report the pass@3 results or remove the sentence.
[Table 2, §3.8] The notation is inconsistent: L is defined as the space of natural language queries, yet Eq. (2) uses J ∈ L for justifications, and Section 3.8 refers to E as "retrieved evidence" even though E was defined as extracted entities in Section 3.2 and C denotes retrieved chunks.
[§3.3] The embedding model is described only as "OpenAI text embedding model"; please specify the exact model name and version for reproducibility.
[§6.1.1] The perplexity score of 57.05 is reported without specifying which language model computed it; please state the model and tokenizer used.
[Table 1] The double check marks ("✓✓") in the AURA row are unexplained; use single check marks or add a note explaining the convention.

Circularity Check

1 steps flagged · score 6.0 of 10

Answer-in-KB leakage lets AURA read the ground-truth actor off retrieved reports, so the headline accuracy is partly a retrieval lookup rather than a derived attribution effort.

renaming known result [Sections 3.3, 3.7, 4.1, and 6.2.1, including Eq. (1)]
"Specifically, 2,199 reports are used to populate the vector database that serves as AURA’s knowledge base (Step 2 in Fig 3), while the remaining 30 reports are reserved as a held-out test set (Step6 in Fig 3). ... Public threat intelligence by Cyfirma [24] directly links this campaign to APT36."

According to Eq. (1), the predicted actor is A = f_attr(Q', E, C), where C is the set of top-k retrieved chunks. The Section 4.1 split removes only the 30 exact test reports from the vector database; it does not remove other reports from the same GitHub collections that explicitly name the ground-truth actor for the same campaign. The case study in Section 6.2.1 shows the mechanism: AURA's own justification cites a Cyfirma report that 'directly links this campaign to APT36,' meaning the retrieved context already contains the answer label. The Semantic Retriever (Section 3.3) has no filter for chunks that already state the attribution, and the Decision Agent only checks relevance.

full rationale

The paper's central numerical claim is the attribution accuracy on 30 held-out reports (Section 5). The held-out construction in Section 4.1 isolates the test reports from LLM pretraining, but not from AURA's own retrieval knowledge base: the 2,199-document corpus is drawn from the same GitHub collections (APTNotes, CyberMonitor) and can contain other vendor reports that name the actor for the same campaign. Since the pipeline is A = f_attr(Q', E, C) and the retrieved context C is fed directly to the Attribution Agent, the prediction is not independent of an input that may contain the answer. The Section 6.2.1 case study confirms this: AURA's justification relies on a Cyfirma report that 'directly links this campaign to APT36,' so the output is read off the retrieved text rather than derived from behavioral indicators. This makes the headline evaluation partially circular: the reported accuracy can be achieved by lookup, and the multi-agent reasoning contribution is not isolated. A secondary weakness is that the LLM-as-Judge evaluation (Section 6.1.2) uses gpt-4o to judge justifications produced by the same model family, but this affects only the justification-quality claim, not the attribution-accuracy numbers. The finding is partial, not a full collapse: if the retrievable corpus lacked explicit labels, the framework could still add value, but the paper does not demonstrate that.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the correctness of test labels, the untested post-cutoff assumption for the held-out reports, the validity of LLM-based extraction and evaluation, and the relevance of vector retrieval. No new physical or conceptual entities are postulated.

free parameters (2)

Retrieval top-k
Number of chunks retrieved per query is not reported; it is a hand-chosen hyperparameter that directly affects the context given to the attribution agent.
Chunk overlap = 50 tokens
Set to 50 tokens to preserve semantic flow between chunks (Section 4.1); hand-chosen and not varied in experiments.

assumptions (5)

domain assumption Ground truth actor labels in the test reports are correct.
The paper uses the actor assignment stated in each public threat report as the target label without independent verification.
ad hoc to paper The 30 test reports are outside the pretraining data of gpt-4o, gpt-4o-mini, Claude 3.5 Haiku, and Claude 3.5 Sonnet.
The paper splits by LLM knowledge cutoff dates (Section 4.1) but does not verify this property for each model; if false, accuracy is inflated by memorization.
domain assumption LLM-as-Judge scores (using gpt-4o) approximate human expert judgment of justification quality.
Used in Section 6.1.2 to support the 'expert-aligned justifications' claim without any human evaluation.
domain assumption Cosine similarity over OpenAI embeddings retrieves context relevant to attribution.
The Semantic Retriever (Section 3.3) assumes that high embedding similarity corresponds to evidence useful for attribution.
domain assumption gpt-4o accurately extracts structured threat artifacts (TTPs, IoCs, tools, timelines) from test reports.
Test inputs are generated by gpt-4o extraction (Section 4.1); extraction errors propagate to the attribution step.

how reviews work

0 comments

Cite this review

Pith. "Pith review of AURA: A Multi-Agent Intelligence Framework for Knowledge-Enhanced Cyber Threat Attribution." pith.science (2026). https://pith.science/paper/SLQN7ASM

@misc{pith2026250610175,
  author       = {Pith},
  title        = {Pith review of: AURA: A Multi-Agent Intelligence Framework for Knowledge-Enhanced Cyber Threat Attribution},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SLQN7ASM}},
  note         = {Machine review of arXiv:2506.10175}
}

read the original abstract

Effective attribution of Advanced Persistent Threats (APTs) increasingly hinges on the ability to correlate behavioral patterns and reason over complex, varied threat intelligence artifacts. We present AURA (Attribution Using Retrieval-Augmented Agents), a multi-agent, knowledge-enhanced framework for automated and interpretable APT attribution. AURA ingests diverse threat data including Tactics, Techniques, and Procedures (TTPs), Indicators of Compromise (IoCs), malware details, adversarial tools, and temporal information, which are processed through a network of collaborative agents. These agents are designed for intelligent query rewriting, context-enriched retrieval from structured threat knowledge bases, and natural language justification of attribution decisions. By combining Retrieval-Augmented Generation (RAG) with Large Language Models (LLMs), AURA enables contextual linking of threat behaviors to known APT groups and supports traceable reasoning across multiple attack phases. Experiments on recent APT campaigns demonstrate AURA's high attribution consistency, expert-aligned justifications, and scalability. This work establishes AURA as a promising direction for advancing transparent, data-driven, and scalable threat attribution using multi-agent intelligence.

Figures

Figures reproduced from arXiv: 2506.10175 by the authors.

**Figure 2.** Architecture of the AURA-based chat-bot system. [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗

**Figure 3.** Dataset Preparation 4. Experiments Setup 4.1. Dataset Effective retrieval augmentation requires a substantial repository of taskspecific knowledge to support the agents during attribution. To build this knowledge base, we collect threat analysis reports published by reputable cybersecurity firms such as Google, CrowdStrike, Kaspersky, and others. The dataset is sourced from publicly available repositories on GitHub… view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: Attribution accuracy of four LLMs across group-wise and nation-wise levels [PITH_FULL_IMAGE:figures/full_fig_p016_4.png]

**Figure 5.** Figure 5: Evaluation of justification quality using four linguistic and semantic measures. [PITH_FULL_IMAGE:figures/full_fig_p018_5.png]

**Figure 6.** Figure 6: Combined LLM-as-Judge evaluation of justifications across four dimensions [PITH_FULL_IMAGE:figures/full_fig_p020_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 31 canonical work pages

[24]

Rani, N., Saha, B., Maurya, V., and Shukla, S. K. Chasing the shadows: Ttps in action to attribute advanced persistent threats. arXiv preprint arXiv:2409.16400\/ (2024)

work page Pith review arXiv 2024
[1]

write newline

" write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...
[2]

write newline

" write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...
[3]

write newline

" write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...
[4]

https://attack.mitre.org/groups/G0037/

Apt36 — mitre att&ck. https://attack.mitre.org/groups/G0037/. Accessed: 2025-05-27

work page 2025
[5]

Aptnotes data: Curated threat intelligence reports on apt activities

APTNotes Contributors . Aptnotes data: Curated threat intelligence reports on apt activities. https://github.com/aptnotes/data, 2025. Accessed: 2025-06-10

work page 2025
[6]

B., Alptekin, H., and C etin, O

B \"o ge, E., Ertan, M. B., Alptekin, H., and C etin, O. Unveiling cyber threat actors: a hybrid deep learning approach for behavior-based attribution. Digital Threats: Research and Practice 6 , 1 (2025), 1--20

work page 2025
[7]

Chen, M., Tworek, J., Jun, H., Yuan, Q., Pinto, H. P. D. O., Kaplan, J., Edwards, H., Burda, Y., Joseph, N., Brockman, G., et al. Evaluating large language models trained on code. arXiv preprint arXiv:2107.03374\/ (2021)

arXiv 2021

Show all 39 references

[8]

Towards effective identification of attack techniques in cyber threat intelligence reports using large language models

Cuong Nguyen, H., Tariq, S., Baruwal Chhetri, M., and Quoc Vo, B. Towards effective identification of attack techniques in cyber threat intelligence reports using large language models. In Companion Proceedings of the ACM on Web Conference 2025\/ (2025), pp. 942--946

2025
[9]

Apt cybercriminal campaign collections

CyberMonitor . Apt cybercriminal campaign collections. https://github.com/CyberMonitor/APT_CyberCriminal_Campagin_Collections, 2025. Accessed: 2025-06-10

2025
[10]

J., and Smeets, M

Egloff, F. J., and Smeets, M. Publicly attributing cyber attacks: a framework. Journal of Strategic Studies 46 , 3 (2023), 502--533

2023
[11]

Identifying authorship in malicious binaries: Features, challenges & datasets

Gray, J., Sgandurra, D., Cavallaro, L., and Blasco Alis, J. Identifying authorship in malicious binaries: Features, challenges & datasets. ACM Computing Surveys 56 , 8 (2024), 1--36

2024
[12]

A survey on llm-as-a-judge

Gu, J., Jiang, X., Shi, Z., Tan, H., Zhai, X., Xu, C., Li, W., Shen, Y., Ma, S., Liu, H., et al. A survey on llm-as-a-judge. arXiv preprint arXiv:2411.15594\/ (2024)

2024 arXiv
[13]

V., Wiest, O., and Zhang, X

Guo, T., Chen, X., Wang, Y., Chang, R., Pei, S., Chawla, N. V., Wiest, O., and Zhang, X. Large language model based multi-agents: A survey of progress and challenges. arXiv preprint arXiv:2402.01680\/ (2024)

2024 arXiv
[14]

J., and Kochenderfer, M

Guru, K., Moss, R. J., and Kochenderfer, M. J. On technique identification and threat-actor attribution using llms and embedding models. arXiv preprint arXiv:2505.11547\/ (2025)

2025 arXiv
[15]

H.-K., Chen, C.-H., Liao, Z.-J., et al

Huang, Y.-T., Vaitheeshwari, R., Chen, M.-C., Lin, Y.-D., Hwang, R.-H., Lin, P.-C., Lai, Y.-C., Wu, E. H.-K., Chen, C.-H., Liao, Z.-J., et al. Mitretrieval: Retrieving mitre techniques from unstructured threat reports by fusion of deep learning and ontology. IEEE Transactions ...

2024
[16]

Irshad, E., and Siddiqui, A. B. Context-aware cyber-threat attribution based on hybrid features. ICT Express\/ (2024)

2024
[17]

Kulal, S., Pasupat, P., Chandra, K., Lee, M., Padon, O., Aiken, A., and Liang, P. S. Spoc: Search-based pseudocode to code. Advances in Neural Information Processing Systems 32\/ (2019)

2019
[18]

A review of attribution technical for apt attacks

Mei, Y., Han, W., Li, S., Wu, X., Lin, K., and Qi, Y. A review of attribution technical for apt attacks. In 2022 7th IEEE International Conference on Data Science in Cyberspace (DSC)\/ (2022), IEEE, pp. 512--518

2022
[19]

Deep learning for threat actor attribution from threat reports

Naveen, S., Puzis, R., and Angappan, K. Deep learning for threat actor attribution from threat reports. In 2020 4th international conference on computer, communication and signal processing (ICCCSP)\/ (2020), IEEE, pp. 1--6

2020
[20]

Noor, U., Anwar, Z., Amjad, T., and Choo, K.-K. R. A machine learning-based fintech cyber threat attribution framework using high-level indicators of compromise. Future Generation Computer Systems 96\/ (2019), 227--242

2019
[21]

No-doubt: Attack attribution based on threat intelligence reports

Perry, L., Shapira, B., and Puzis, R. No-doubt: Attack attribution based on threat intelligence reports. In 2019 IEEE International Conference on Intelligence and Security Informatics (ISI)\/ (2019), IEEE, pp. 80--85

2019
[22]

Rani, N., Saha, B., Kumar, R., and Shukla, S. K. Genesis of cyber threats: Towards malware-based advanced persistent threat (apt) attribution. In 2024 IEEE 6th International Conference on Trust, Privacy and Security in Intelligent Systems, and Applications (TPS-ISA)\/ (2024), ...

2024
[23]

Rani, N., Saha, B., Maurya, V., and Shukla, S. K. Ttphunter: Automated extraction of actionable intelligence as ttps from narrative threat reports. In Proceedings of the 2023 Australasian Computer Science Week . 2023, pp. 126--134

2023
[25]

Rani, N., Saha, B., Maurya, V., and Shukla, S. K. Ttpxhunter: Actionable threat intelligence extraction as ttps from finished cyber threat reports. Digital Threats: Research and Practice 5 , 4 (2024), 1--19

2024
[26]

Rani, N., Saha, B., and Shukla, S. K. A comprehensive survey of automated advanced persistent threat attribution: Taxonomy, methods, challenges and open research problems. Journal of Information Security and Applications 92\/ (2025), 104076

2025
[27]

Turning aid into attack: Exploitation of pakistan’s youth laptop scheme to target india

Research, C. Turning aid into attack: Exploitation of pakistan’s youth laptop scheme to target india. https://www.cyfirma.com/research/turning-aid-into-attack-exploitation-of-pakistans-youth-laptop-scheme-to-target-india/, 2025. Accessed: 2025-05-27

2025
[28]

why should i trust you?

Ribeiro, M. T., Singh, S., and Guestrin, C. " why should i trust you?" explaining the predictions of any classifier. In Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining\/ (2016), pp. 1135--1144

2016
[29]

Attributing cyber attacks

Rid, T., and Buchanan, B. Attributing cyber attacks. Journal of Strategic Studies 38 , 1-2 (2015), 4--37

2015
[30]

Deepapt: nation-state apt attribution using end-to-end deep neural networks

Rosenberg, I., Sicard, G., and David, E. Deepapt: nation-state apt attribution using end-to-end deep neural networks. In Artificial Neural Networks and Machine Learning--ICANN 2017: 26th International Conference on Artificial Neural Networks, Alghero, Italy, September 11-14, 2...

2017
[31]

Saha, B., Rani, N., and Shukla, S. K. Malaware: Automating the comprehension of malicious software behaviours using large language models (llms). arXiv preprint arXiv:2504.01145\/ (2025)

2025 arXiv
[32]

Saha, B., and Shukla, S. K. Malgen: A generative agent framework for modeling malicious software in cybersecurity, 2025

2025
[33]

Under false flag: using technical artifacts for cyber attack attribution

Skopik, F., and Pahi, T. Under false flag: using technical artifacts for cyber attack attribution. Cybersecurity 3\/ (2020), 1--20

2020
[34]

Attribution of Advanced Persistent Threats

Steffens, T. Attribution of Advanced Persistent Threats . Springer, 2020

2020
[35]

Pawn storm uses brute force and stealth against high-value targets

Trend Micro Research . Pawn storm uses brute force and stealth against high-value targets. https://www.trendmicro.com/en_in/research/24/a/pawn-storm-uses-brute-force-and-stealth.html, 2024. Accessed: 2025-06-09

2024
[36]

Flesch–kincaid readability tests

Wikipedia contributors . Flesch–kincaid readability tests. https://en.wikipedia.org/wiki/Flesch Accessed: 2025-06-11

2025
[37]

Autogen: Enabling next-gen llm applications via multi-agent conversation

Wu, Q., Bansal, G., Zhang, J., Wu, Y., Li, B., Zhu, E., Jiang, L., Zhang, X., Zhang, S., Liu, J., et al. Autogen: Enabling next-gen llm applications via multi-agent conversation. arXiv preprint arXiv:2308.08155\/ (2023)

2023 arXiv
[38]

Judging llm-as-a-judge with mt-bench and chatbot arena

Zheng, L., Chiang, W.-L., Sheng, Y., Zhuang, S., Wu, Z., Zhuang, Y., Lin, Z., Li, Z., Li, D., Xing, E., et al. Judging llm-as-a-judge with mt-bench and chatbot arena. Advances in Neural Information Processing Systems 36\/ (2023), 46595--46623

2023
[39]

Naver ending game: Lazarus apt targeting south korean users

Zscaler ThreatLabz . Naver ending game: Lazarus apt targeting south korean users. https://www.zscaler.com/blogs/security-research/naver-ending-game-lazarus-apt, 2024. Accessed: 2025-06-09

2024

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [24]

Rani, N., Saha, B., Maurya, V., and Shukla, S. K. Chasing the shadows: Ttps in action to attribute advanced persistent threats. arXiv preprint arXiv:2409.16400\/ (2024)

work page Pith review arXiv 2024

[2] [1]

write newline

" write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

[3] [2]

write newline

" write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

[4] [3]

write newline

" write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

[5] [4]

https://attack.mitre.org/groups/G0037/

Apt36 — mitre att&ck. https://attack.mitre.org/groups/G0037/. Accessed: 2025-05-27

work page 2025

[6] [5]

Aptnotes data: Curated threat intelligence reports on apt activities

APTNotes Contributors . Aptnotes data: Curated threat intelligence reports on apt activities. https://github.com/aptnotes/data, 2025. Accessed: 2025-06-10

work page 2025

[7] [6]

B., Alptekin, H., and C etin, O

B \"o ge, E., Ertan, M. B., Alptekin, H., and C etin, O. Unveiling cyber threat actors: a hybrid deep learning approach for behavior-based attribution. Digital Threats: Research and Practice 6 , 1 (2025), 1--20

work page 2025

[8] [7]

Chen, M., Tworek, J., Jun, H., Yuan, Q., Pinto, H. P. D. O., Kaplan, J., Edwards, H., Burda, Y., Joseph, N., Brockman, G., et al. Evaluating large language models trained on code. arXiv preprint arXiv:2107.03374\/ (2021)

arXiv 2021