Skip to content

Scan output (JSON)

agentgate scan --format json emits a single JSON document:

{
"$schema": "https://agentgate-1ep.pages.dev/schemas/scan-report.v1.json",
"reportVersion": 1,
"agentgateVersion": "0.1.0",
"generatedAt": "2026-08-03T12:00:00Z",
"target": { "kind": "auto-discovery", "paths": ["~/.claude.json"] },
"summary": { "critical": 1, "high": 2, "medium": 0, "low": 1, "total": 4 },
"findings": [
{
"id": "F-0001",
"rule": "advisory-match",
"category": "supply-chain",
"severity": "critical",
"title": "mcp-remote 0.1.10 matches advisory MCPA-2025-0001 (CVE-2025-6514)",
"description": "OS command injection when connecting to untrusted MCP servers…",
"server": {
"name": "remote-proxy",
"package": "mcp-remote",
"version": "0.1.10",
"ecosystem": "npm",
"transport": "stdio"
},
"evidence": { "configPath": "~/.claude.json", "snippet": "" },
"advisoryIds": ["MCPA-2025-0001"],
"remediation": "Upgrade to mcp-remote >= 0.1.16."
}
]
}
Field Required Description
reportVersion yes Integer format version (this page: 1).
agentgateVersion yes CLI version.
generatedAt yes ISO-8601 UTC.
target yes What was scanned: kind (auto-discovery | config | package), paths.
summary yes Finding counts by severity plus total. Must equal the aggregate of findings.
findings[] yes May be empty.
Field Required Description
id yes Stable within a report (F-nnnn).
rule yes Machine ID of the rule that fired (e.g. advisory-match, hidden-unicode).
category yes One of the shared categories: tool-poisoning, credential-leak, overprivileged, auth-missing, ssrf, rce-vectors, supply-chain.
severity yes critical | high | medium | low.
title yes One line, human-readable.
description yes Full explanation (Markdown allowed).
server yes name required; package, version, ecosystem, transport when known.
evidence no Rule-specific: configPath, toolName, snippet, …
advisoryIds no Matching advisory database IDs.
remediation no Actionable fix.
  • The report viewer renders any conforming document client-side (nothing is uploaded).
  • SARIF output (--format sarif) is a lossy projection of the same findings for GitHub code scanning.