Ars Deep Research
Security checks across static analysis, malware telemetry, and agentic risk
Overview
This is a coherent instruction-only academic research workflow; the main things to notice are broad research delegation, web/file use, optional cross-model sharing, and limited provenance metadata.
This skill appears safe for normal academic research use, but treat it as a broad research assistant: do not provide confidential files or private corpus notes unless needed, verify citations and source claims, and check the optional cross-model setting and package provenance before using it for sensitive work.
Static analysis
No static analysis findings were reported for this release.
VirusTotal
VirusTotal findings are pending for this skill version.
Risk analysis
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Your research topic, source lists, and selected file context may be passed through several delegated research agents and used in web-enabled research steps.
The skill delegates research phases to subagents with file and web tool access. This is expected for deep research, but users should know the workflow can browse and use file-context during research tasks.
delegate_task(... toolsets=["file", "web"])
Use the skill for intended research tasks, avoid providing unnecessary sensitive documents, and review generated citations and claims before relying on them.
It may be harder to independently verify that this package exactly matches the claimed upstream academic-research skill adaptation.
The registry does not provide a source URL or homepage for this packaged skill, even though the files include attribution to an upstream repository.
Source: unknown Homepage: none
If provenance matters, compare the installed files against the cited upstream repository and confirm the publisher/version before relying on it.
Private citation notes or unpublished research metadata could enter the skill’s reasoning context if you include them in a corpus.
The bibliography agent can read user-provided literature corpus metadata, including optional notes, and use it as research context.
Read the five required fields and any optional fields present (`venue`, `doi`, `tags`, `abstract`, `user_notes`).
Remove private notes or sensitive metadata from literature corpora unless you are comfortable having them processed for the research task.
If this optional setting is enabled, draft research material may be shared outside the main agent/model boundary.
The skill documents an optional cross-model verification path that sends reviewed material to another model/API when enabled.
When `ARS_CROSS_MODEL` is set, after completing each checkpoint report, send the reviewed material ... to the cross-model for an independent critique.
Only enable cross-model review when the provider and data handling are acceptable for your material, especially for confidential or unpublished work.