Deep Research with MCP
AdvisoryAudited by Static analysis on Apr 30, 2026.
Overview
No suspicious patterns detected.
Findings (0)
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.
The skill may not work as expected unless those tools and MCP servers are installed and trusted.
The skill depends on local binaries and preconfigured MCP search servers. This is consistent with the research purpose, but users should verify those external components, especially because the registry metadata did not declare required binaries.
requires: bins: ["mcporter", "curl", "python3"] ... mcp_requirements: ... "MiniMax.web_search" ... "web-search-prime.web_search_prime"
Confirm the MCP servers, mcporter, curl, and python3 are installed from trusted sources before using the skill.
The agent will make external search and web requests while researching.
The workflow instructs the agent to call MCP tools and use local commands to fetch and process web pages. These commands are central to the research function and are not hidden.
mcporter call MiniMax.web_search query:"<sub-question>" ... curl -sL "<url>" | python3 -c
Use the skill for topics you are comfortable sending to configured search providers, and review generated reports and saved files.
Research topics and possibly user-provided context may be visible to the configured MCP search provider.
Research queries are routed through configured MCP search servers. This is expected for the skill, but the privacy and identity boundaries depend on those external MCP providers.
Uses MCP-configured search tools (minimax web_search or zai-mcp-server web_search_prime).
Check the privacy and access policies of the MCP search tools before using the skill for confidential research.
If used, a separate agent session may perform the searches and write the report before returning results.
The skill documents optional delegation to a spawned sub-agent. The delegation is disclosed and scoped to the research task, but users should be aware of it.
For Sub-Agent Usage ... sessions_spawn( task: "Run deep research on [TOPIC]... Save report... wake the main session", label: "research-[slug]", model: "opus" )
Only use the sub-agent workflow when you intentionally want delegated or background research, and confirm the output path.