ArXiv Watcher
v1.0.0Search and summarize papers from ArXiv. Use when the user asks for the latest research, specific topics on ArXiv, or a daily summary of AI papers.
⭐ 9· 9.7k·111 current·117 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
high confidencePurpose & Capability
The included shell script calls the official ArXiv API and the SKILL.md capabilities (search, summarize, deep dive) match the declared purpose. No unrelated binaries, services, or credentials are requested.
Instruction Scope
SKILL.md contains a MANDATORY instruction to append every discussed paper to memory/RESEARCH_LOG.md. This forces persistent recording of user-facing summaries (and any extracted PDF content) without asking the user or giving an opt-out. It also tells the agent to use web_fetch on PDFs (external content retrieval) which is expected for 'deep dive' but increases the blast radius and privacy exposures.
Install Mechanism
No install spec and only a tiny helper script are present. The script uses curl to call export.arxiv.org — this is proportionate and low-risk from an install perspective.
Credentials
No environment variables, credentials, or config paths are requested. The skill does not ask for unrelated secrets or broad filesystem access beyond writing to its own memory file.
Persistence & Privilege
Although always:false and no elevated platform privileges are requested, the skill explicitly directs persistent storage to memory/RESEARCH_LOG.md on every use. That mandatory persistence is a privacy and data-governance concern and should be optional or permissioned.
What to consider before installing
This skill appears to perform legitimate ArXiv searches, but it will automatically append summaries to memory/RESEARCH_LOG.md every time it discusses a paper and may fetch full PDFs. Before installing, consider: (1) Do you want the agent to persistently log all papers to memory by default? If not, modify SKILL.md to make saving optional or require user confirmation. (2) Fetching PDFs retrieves external files — ensure you’re comfortable with the agent downloading and processing those documents. (3) The helper script does not URL-encode user queries; you may want to sanitize inputs. If you accept persistent logging and external fetches, the skill is coherent; otherwise treat it as a privacy risk and either edit the skill to remove the mandatory write or decline installation.Like a lobster shell, security has layers — review code before you run it.
latestvk9700rxx0nf0jc0tw9nrjm3g057zz8ry
License
MIT-0
Free to use, modify, and redistribute. No attribution required.