ResearchMonitor
PassAudited by ClawScan on May 1, 2026.
Overview
This skill appears benign: it monitors research topics using web search and local history, but users should know their topics and seen items are stored locally and used in external searches.
Before installing, check that the topics in research_config.json are appropriate to store locally and search externally. The included helper script appears scoped to managing that local configuration and does not show credential use, exfiltration, or destructive behavior.
Findings (2)
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.
If a configured topic reveals private research plans or sensitive interests, it may appear in external search queries.
The skill directs the agent to send configured research topics to a web-search tool. This is central to the stated monitoring purpose, but it means topic text may be exposed to search providers.
For each topic, use `search_web` to look for: ... "new research papers [topic] [current month/year]" ... "upcoming conferences [topic] [current year]"
Use non-sensitive topic names when possible and review research_config.json before letting the skill perform broad searches.
Stored state can reveal research interests and can affect future reports; stale or edited entries could cause the agent to skip or repeat items.
The helper script persists seen item identifiers, and the skill also uses stored topics and last_checked to influence future runs.
config.setdefault('seen_items', []).append(identifier)
save_config(config)Periodically review or clear research_config.json if topics or seen-item history should not persist.