Deepscan Monitor
PassAudited by ClawScan on May 1, 2026.
Overview
This is a coherent instruction-only skill for managing PapersFlow DeepScan runs, with user-noticeable use of external DeepScan tools and stored report history.
This skill appears safe to install as an instruction-only DeepScan helper. Before using it, be explicit about whether you want to start a new run or inspect an existing one, and narrow any request that uses stored DeepScan history to the reports or time period you intend.
Findings (3)
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.
A user request could result in a new long-running DeepScan job being started and monitored.
The skill instructs the agent to start and repeatedly poll an asynchronous external research job. This is aligned with the skill purpose, but starting long-running work is a meaningful action the user should intend.
1. Use `run_deepscan` to start the job. 2. Immediately tell the user that the run is asynchronous. 3. Poll with `get_deepscan_live_snapshot`
Before starting a new run, make the research question and intended action clear; use status/report tools instead when the user only wants an existing run checked.
Prior DeepScan reports may be reused in later summaries, which could include sensitive or unrelated research context if the request is broad.
The skill may use stored DeepScan history across prior reports. This is purpose-aligned when requested, but it means past report content can influence new summaries.
Use `summarize_evidence` when the user wants a cross-report summary from stored DeepScan history.
Specify which reports, topics, or time range should be included when asking for cross-report summaries.
Research prompts, progress data, and report outputs may pass through the configured DeepScan/MCP service.
The skill references an MCP server and DeepScan tools, indicating communication with an external tool/provider boundary. This is disclosed and central to the purpose.
Do not imply the MCP server will push completion notifications into Claude automatically.
Use this only with the intended PapersFlow DeepScan connection and avoid submitting research topics or report content that should not be shared with that service.