io.github.cyanheads/pentest-mcp-server

The only declared environment variable is MCP_LOG_LEVEL, which is typically a logging-level setting rather than a sensitive credential. There is no evidence of API keys, account passwords, or high-value tokens being required, so credential-abuse risk appears low.

The tool is declared to access pentest.caseyjhand.com, so there is network egress. The materials do not specify what user data may be transmitted, but connecting to its stated endpoint is a normal capability for this type of tool and warrants monitoring of outbound traffic scope and contents.

The system marks this tool as executes-code, indicating it can execute code or spawn processes locally. Given its positioning as a pentest methodology engine, this is expected functionality, but it still means it should run in an isolated environment with limited host privileges.

The available materials do not specify which files, directories, or other resources it can read or write. As an MCP tool that can execute code, it should be assumed to potentially access local workspace or testing data. There is no explicit evidence of permissions far beyond its stated function, but transparency is limited.

Positive factors include the official registry source, an auditable open-source repository, and updates within the past year, all of which reduce supply-chain risk. However, the lack of a README, no declared license, and very low community adoption (0 stars) limit verifiability and maturity, so caution is appropriate rather than a high-risk rating.