REPOSITORY EVIDENCE / INDEXABLE REPORT

modelcontextprotocol/inspector — MCP & Agent Security Evidence Report

MCP configuration, repository automation and MCP v2 dependency evidence were observed in this selected bounded review. Evidence observations are review signals, not a safe/unsafe verdict.

Is modelcontextprotocol/inspector safe to run?

ShadowMCP does not issue a safe/unsafe verdict. This report records bounded evidence tied to a specific commit so reviewers can identify what deserves inspection before use.

Scan identity

Report schemashadowmcp.repo-evidence.v9
Scannershadowmcp.web-repo-scan.v22 semantics
Scan completed2026-09-25T08:31:00.000Z
Repository branchmain
Scanned commit1e31c78fbf81a989e8eb47021c6281d7876ad7fd
Indexation gateIndexable quality gate passed.

Evidence summary

Review priority
Priority review

This is a triage priority, not a repository risk rating.

This is the official visual testing tool for MCP servers. The bounded review observed a root MCP configuration pointing at MCP documentation, extensive validation/build automation and explicit Model Context Protocol v2 client/server dependencies.

Evidence categories

MCP configuration

Found

Selected configuration evidence was observed. This does not establish how any user's client is configured.

  • .mcp.jsonmcpServers entry points to the official MCP documentation HTTP endpoint

Shell / command execution

Not established

No shell/command execution claim is made from the selected bounded evidence.

CI/CD automation

Found

Repository automation and validation tooling are present in the selected scope.

  • .github/GitHub automation directory observed
  • package.jsonvalidation, coverage, build-gate and smoke-test scripts declared

Supply-chain / dependencies

Found

Selected package dependency declarations were observed; this is not a complete dependency audit.

  • package.json@modelcontextprotocol/client, core, server and server-legacy 2.0.0 dependencies declared

Scope and limitations

Scan mode: BOUNDED_ROOT_AND_SELECTED_DEPTH1_PLUS_TARGETED_SEARCH. General recursive crawl: NO.

  • Public, selected and bounded repository evidence only.
  • FOUND means evidence observed, not risk or malicious behavior proven.
  • Evidence was reviewed against immutable commit 1e31c78fbf81a989e8eb47021c6281d7876ad7fd.
  • No repository clone or target-code execution.
  • No secret/token values are included.
  • No CVE or package-reputation lookup.
  • No numeric security score and no safe/unsafe verdict.

Repository context

Trust/community metadata is descriptive context only; it is not a security guarantee.

Repository age / activitycreated 2024-10-03 · active 2026-09-25
Community signalsstars 10,952 · forks 1,528 · open issues 50
OwnershipOrganization · modelcontextprotocol
LicenseOther / NOASSERTION in GitHub metadata
Repository statearchived NO · fork NO · default branch main

Continue the review