Atlas / MCP servers / zoharbabin / Google Researcher

Google ResearcherSAFE

mcp/zoharbabin/google-researcher-1

⚠️ DEPRECATED — Use https://github.com/zoharbabin/web-researcher-mcp instead

Verdict
SAFE
Grade
B
Trust score
87 /100
Exposed tools
12 12r · 0w · 0d
Transport
stdio · streamable-http
License
MIT
Stars
36
01

Overview

From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.

⚠️ This project has been superseded by [`web-researcher-mcp`](https://github.com/zoharbabin/web-researcher-mcp). A complete rewrite in Go: same tools, but it cites real sources and stays honest — citation verification, retraction flagging, eight search providers behind one interface, and a single signed static binary. No Node.js, no npm.

Migrate in one line

Your API keys and tool calls don't change — only how the server is launched. Pick whichever you prefer (no Node required):

# macOS / Linux — installs the binary and registers it with Claude Code automatically:
curl -fsSL https://raw.githubusercontent.com/zoharbabin/web-researcher-mcp/main/install.sh | sh

# Windows (PowerShell):
powershell -ExecutionPolicy Bypass -c "irm https://raw.githubusercontent.com/zoharbabin/web-researcher-mcp/main/install.ps1 | iex"

# Python users (uv) — run with no install:
uvx web-researcher-mcp

Before / after (MCP client config)

The config keeps the same shape — swap the command/args, keep your env:

// BEFORE — google-researcher-mcp (npx / Node)
{
"mcpServers": {
"google-researcher": {
"command": "npx",
"args": ["-y", "google-researcher-mcp"],
"env": { "GOOGLE_CUSTOM_SEARCH_API_KEY": "...", "GOOGLE_CUSTOM_SEARCH_ID": "..." }
}
}
}

// AFTER — web-researcher-mcp (installed binary; or use "uvx" + args ["web-researcher-mcp"])
{
"mcpServers": {
"web-researcher": {
"command": "web-researcher-mcp",
"env": { "GOOGLE_CUSTOM_SEARCH_API_KEY": "...", "GOOGLE_CUSTOM_SEARCH_ID": "..." }
}
}
}
Don't run both at once — they register identical tool names. Remove the google-researcher entry before adding web-researcher.

→ [Full migration guide](https://github.com/zoharbabin/web-researcher-mcp/blob/main/docs/MIGRATION.md)

Your existing GOOGLE_CUSTOM_SEARCH_API_KEY and GOOGLE_CUSTOM_SEARCH_ID work unchanged.

Why the Rew

Read from source at commit 89aae67902daOBSERVED · 2026-10-08
02

Connect

Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.

claude-code (npm)
claude mcp add google-researcher-mcp --env GOOGLE_CUSTOM_SEARCH_API_KEY=${GOOGLE_CUSTOM_SEARCH_API_KEY} -- npx -y [email protected]
03

Exposed tools (12)

12 read · 0 write · 0 destructive.

ToolRiskDescription
competitive-analysisreadCompare multiple companies or products across key dimensions
comprehensive-researchreadResearch a topic thoroughly using multiple sources, with synthesis and citations
due-diligence-backgroundreadBackground & reputational due diligence research on a company: automatically discovers subsidiaries, prior names, executives, then screens for security breaches, compliance violations, lawsuits, executive scandals, and security posture
fact-checkreadVerify a claim using multiple authoritative sources
google_searchread
literature-reviewreadConduct an academic literature review with proper citations
news-briefingreadGet a current news summary on a topic
patent-portfolio-analysisreadAnalyze a company\
scrape_pageread
search_and_scraperead
summarize-urlreadExtract and summarize content from a specific URL
technical-deep-divereadIn-depth technical investigation of a technology or concept
04

Trust audit

SAFEgrade B · trust 87/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

LayerWhat it checksResult
L0Provenance & inventoryPASS
L1Static analysis of the codePASS
L2Instruction surface (what it tells the agent)PASS
L3Class-specific surfacePASS
L4Behavioural (sandbox)SKIPPED

What the source does

Filesystem
declared (3 observation(s))
Network
declared (6 observation(s))
Shell
none-observed
Dependencies
not all pinned
Secrets in source
none-found

Findings (17)

LOWInventory / provenance · inv.hidden_file · CWE-1104
.mcpbignore
.mcpbignore
Why it matters. hidden member outside the usual dotfiles
Fix. review its purpose
LOWFilesystem / path · fs.traversal · CWE-22, CWE-59
src/shared/oauthScopesDocumentation.ts:117
- [MCP Server README — OAuth 2.1 Authorization](../../README.md#oauth-21-authorization)
LOWFilesystem / path · fs.traversal · CWE-22, CWE-59
tests/performance/benchmark.spec.ts:29
import { PersistentCache, HybridPersistenceStrategy } from '../../src/cache/index.js';
LOWFilesystem / path · fs.traversal · CWE-22, CWE-59
tests/performance/benchmark.spec.ts:30
import { PersistentEventStore } from '../../src/shared/persistentEventStore.js';
LOWFilesystem / path · fs.traversal · CWE-22, CWE-59
tests/performance/benchmark.spec.ts:200
const { createAppAndHttpTransport } = await import('../../src/server.js');
LOWNetwork egress · net.metadata_ip · CWE-200, CWE-319
src/shared/urlValidator.spec.ts:51
it('should block AWS metadata endpoint (169.254.169.254)', async () => {
Why it matters. cloud metadata endpoint: the classic SSRF credential grab
LOWNetwork egress · net.metadata_ip · CWE-200, CWE-319
src/shared/urlValidator.spec.ts:52
await expect(validateUrlForSSRF('http://169.254.169.254/latest/meta-data/')).rejects.toThrow(SSRFProtectionError);
Why it matters. cloud metadata endpoint: the classic SSRF credential grab
LOWNetwork egress · net.metadata_ip · CWE-200, CWE-319
src/shared/urlValidator.spec.ts:56
await expect(validateUrlForSSRF('http://metadata.google.internal/')).rejects.toThrow(SSRFProtectionError);
Why it matters. cloud metadata endpoint: the classic SSRF credential grab
LOWNetwork egress · net.metadata_ip · CWE-200, CWE-319
src/shared/urlValidator.spec.ts:106
await expect(validateUrlForSSRF('http://169.254.169.254/', opts)).resolves.toBeUndefined();
Why it matters. cloud metadata endpoint: the classic SSRF credential grab
LOWNetwork egress · net.metadata_ip · CWE-200, CWE-319
src/shared/urlValidator.spec.ts:115
validateUrlForSSRF('http://metadata.google.internal/', opts)
Why it matters. cloud metadata endpoint: the classic SSRF credential grab
LOWNetwork egress · net.raw_ip · CWE-200, CWE-319
src/shared/urlValidator.spec.ts:30
await expect(validateUrlForSSRF('http://127.0.0.1/secret')).rejects.toThrow(SSRFProtectionError);
LOWNetwork egress · net.raw_ip · CWE-200, CWE-319
src/shared/urlValidator.spec.ts:34
await expect(validateUrlForSSRF('http://10.0.0.1/internal')).rejects.toThrow(SSRFProtectionError);
LOWNetwork egress · net.raw_ip · CWE-200, CWE-319
src/shared/urlValidator.spec.ts:38
await expect(validateUrlForSSRF('http://192.168.1.1/admin')).rejects.toThrow(SSRFProtectionError);
LOWNetwork egress · net.raw_ip · CWE-200, CWE-319
src/shared/urlValidator.spec.ts:42
await expect(validateUrlForSSRF('http://172.16.0.1/')).rejects.toThrow(SSRFProtectionError);
LOWNetwork egress · net.raw_ip · CWE-200, CWE-319
src/shared/urlValidator.spec.ts:46
await expect(validateUrlForSSRF('http://0.0.0.0/')).rejects.toThrow(SSRFProtectionError);
LOWSupply chain · supply.unpinned · CWE-829, CWE-1357
package.json
@types/node, typescript
Why it matters. 2 dependency range(s) float
Fix. pin exact versions or ship a lockfile
INFOSupply chain · prompt.pipe_to_shell · CWE-829, CWE-1357
README.md:15
curl -fsSL https://raw.githubusercontent.com/zoharbabin/web-researcher-mcp/main/install.sh | sh

Gates applied: no_behavioural_pass.

Audited 2026-10-08 · audit v0.4.1 · source sha 89aae67902dafull audit observations/trust-audit/mcp-server/zoharbabin__google-researcher-1.json · Report an issue / request a re-scan
05

Audit history

Every audit this server has had. A grade with a past is a grade somebody is still checking.

DateSourceVerdictGradeScoreChange
2026-10-0889aae67902daSAFEB87first audit
06

Questions

What is the Google Researcher MCP server?

⚠️ DEPRECATED — Use https://github.com/zoharbabin/web-researcher-mcp instead

What tools does Google Researcher expose?

12 in total: 12 read-only, 0 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.

Is Google Researcher safe to connect to an agent?

The audit found nothing in the source that contradicts what it says it does, and graded it B (87/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B.

What credentials does Google Researcher need?

It reads CACHE_ADMIN_KEY, GOOGLE_CUSTOM_SEARCH_API_KEY, OAUTH_AUDIENCE and OAUTH_ISSUER_URL from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.

How does Google Researcher run?

It speaks stdio and streamable-http, so it runs as a local process your client starts. It is published on npm as google-researcher-mcp at 7.0.0.

How current is this page?

The grade is for one exact copy of the source (89aae67902da), read on 2026-10-08. The repository is watched and re-audited when it changes.

Advertisement