Skillv2.0.0

ClawScan security

OpusFlame Deep Research · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

BenignFeb 25, 2026, 4:34 AM

Verdict: benign
Confidence: medium
Model: gpt-5-mini
Summary: The skill's instructions, requirements, and artifacts are consistent with its stated purpose of running multi-model, framework-driven research; it does not request extra credentials or install code, but it does mandate heavy web fetching and persistent memory writes which you should understand before enabling.
Guidance: This skill appears internally consistent with its stated goal of deep, multi-model research. Before installing, note that it will: (1) perform many automated web searches and fetch full articles (which can hit paywalls, produce heavy network traffic, or surface copyrighted material), (2) save per-model outputs and the merged report to memory/research/ (persisting the query and results), and (3) autonomously spawn parallel sessions to run those model agents. Because the skill comes from an unknown source with no homepage, consider limiting use to non-sensitive topics, avoid providing secrets or proprietary documents as prompts, inspect any saved memory files after a run (and clear them if necessary), and verify that your platform’s web_fetch policies (rate limits, terms of service for target sites) are acceptable. If you need stronger assurance, ask the publisher for provenance or a link to documentation/source code before enabling for general use.

Purpose & Capability: okName and description (autonomous multi-model deep research) match the SKILL.md: it explicitly spawns four model agents, requires many web searches and fetches, applies frameworks, and writes per-model outputs and a merged report. No unrelated credentials, binaries, or installs are requested.
Instruction Scope: noteInstructions are narrowly focused on research tasks but demand extensive web activity (minimum 10 searches per model, 15 unique source URLs per model, web_fetch of full articles) and persistent saves under memory/research/*. This is proportionate for deep research but increases network activity, potential scraping of paywalled or copyrighted material, and the chance that user-provided sensitive topics could be written to memory. The skill does not instruct reading unrelated files or env vars.
Install Mechanism: okInstruction-only skill with no install spec and no code files. Lowest-risk install profile — nothing is written to disk by an installer.
Credentials: okNo environment variables, credentials, or config paths are requested. The listed memory save paths are reasonable given the purpose. There are no disproportionate secret requests.
Persistence & Privilege: okDoes not request always:true or other elevated privileges. It does instruct saving outputs to memory/research/—this is expected for research but means results are persisted and should be reviewed/cleared if sensitive.