Is Keboola MCP server safe?
What Keboola MCP server does
Keboola MCP Server connects AI agents and MCP clients to a Keboola project, exposing storage access, SQL queries, transformations and job triggers as tools. PyPI lists Keboola (devel@keboola.com) as author; it runs locally or as a remote server hosted on each multi-tenant Keboola stack, with OAuth. It calls the Keboola API with your project credentials, or with a browser sign-in session it stores and refreshes. Before installing, consider read-only mode (X-Read-Only-Mode) or an X-Allowed-Tools list, and working in a development branch, where the README says operations are scoped to that branch rather than production.
- Python network egress
- Python filesystem read
- Python filesystem write/delete
What to do: Nothing here argues against installing it. Grant the capabilities it lists only if you expect the tool to need them.
Keboola MCP server is an MCP server analyzed by SkillTotal's deterministic static scanner. The scan found no malicious indicators. It can: delegated authentication, filesystem read, filesystem write, mcp tools detected and network egress — capabilities are what the code can do, not a verdict on intent. Risk score 0/100 (low).
keboola-mcp-server 1.119.0
Get notified when this component changes
Change alerts aren't live yet. Leave your email and we'll let you know when they launch. We use it only for this and delete it on request.
Automated static-analysis result. It can contain false positives and false negatives, and is not a claim about the intent of Keboola MCP server's authors. Report a false positive.
Behavioral traits
How this component maps to the CSA agentic threat model. Descriptive — it never affects the risk score.
Findings (5)
The component reads files from disk.
return cast(dict, json.loads(_CREDENTIALS_PATH.read_text()))
jwt = Path(path).read_text().strip()
return resources.files(__package__).joinpath(name).read_text(encoding='utf-8')
return [(p.name, p.read_text()) for p in files]
_QUERY_SERVICE_QUERY_DATA_FUNCTION_CODE = _DATA_APP_RESOURCES.joinpath('qsapi_query_data_code.py').read_text(
encoding='utf-8'
)_STORAGE_QUERY_DATA_FUNCTION_CODE = _DATA_APP_RESOURCES.joinpath('sapi_query_data_code.py').read_text(encoding='utf-8')Why it matters: Usually legitimate, but worth confirming it can't be steered into reading sensitive files.
Fix: Confirm which files are read and that paths cannot be influenced by untrusted input to reach sensitive locations.
The component writes or deletes files on disk.
with open(self._output_path, mode='w', encoding='utf-8') as f:
Why it matters: Usually legitimate, but worth confirming the paths can't be controlled by untrusted input.
Fix: Confirm which files are written/deleted and that paths cannot be influenced by untrusted input.
The component makes outbound network requests.
import httpx
_AUTH_TIMEOUT = httpx.Timeout(connect=5.0, read=30.0, write=10.0, pool=5.0)
async with httpx.AsyncClient(timeout=_AUTH_TIMEOUT, transport=transport) as client:
async with httpx.AsyncClient(timeout=_AUTH_TIMEOUT, transport=transport) as client:
async with httpx.AsyncClient(timeout=_AUTH_TIMEOUT, transport=transport) as client:
async with httpx.AsyncClient(timeout=_AUTH_TIMEOUT, transport=transport) as client:
async with httpx.AsyncClient(timeout=_AUTH_TIMEOUT, transport=transport) as client:
async with httpx.AsyncClient(timeout=_AUTH_TIMEOUT, transport=transport) as client:
async with httpx.AsyncClient(timeout=_AUTH_TIMEOUT, transport=transport) as client:
import httpx
self._timeout = timeout or httpx.Timeout(connect=5.0, read=30.0, write=10.0, pool=5.0)
self._http = httpx.AsyncClient(timeout=self._timeout, transport=self._transport)
self._timeout = timeout or httpx.Timeout(connect=5.0, read=30.0, write=10.0, pool=5.0)
async with httpx.AsyncClient(timeout=self._timeout, transport=self._transport) as client:
import httpx
self.timeout = timeout or httpx.Timeout(connect=5.0, read=60.0, write=10.0, pool=5.0)
return httpx.AsyncHTTPTransport(retries=2)
raise httpx.HTTPStatusError('\n'.join(message_parts), request=response.request, response=response) from easync with httpx.AsyncClient(timeout=self.timeout, transport=self._create_transport()) as client:
async with httpx.AsyncClient(timeout=self.timeout, transport=self._create_transport()) as client:
async with httpx.AsyncClient(timeout=timeout or self.timeout, transport=transport) as client:
async with httpx.AsyncClient(timeout=self.timeout, transport=self._create_transport(retry=retry)) as client:
async with httpx.AsyncClient(timeout=self.timeout, transport=self._create_transport()) as client:
async with httpx.AsyncClient(timeout=self.timeout, transport=self._create_transport()) as client:
import httpx
Why it matters: Usually legitimate, but confirm the destinations are expected and no sensitive data leaves.
Fix: Confirm the destination hosts are expected and that no sensitive data is sent off-host.
An OAuth 2.0 / OpenID Connect delegated-authentication flow was detected (authorization-code / refresh-token / token-exchange grant, an OIDC authorize/discovery endpoint or id_token, or a delegation library). Tools authenticate with the end user's delegated, scoped credentials rather than a long-lived embedded service credential. (9 occurrence(s) shown as evidence).
/oauth/authorize, so an id anyone could derive would let a user approve a pair out of band, skipping
self._authorize_url = urljoin(server_url, '/oauth/authorize')
Builds the URL that sends the browser to Connection's own `/oauth/authorize` carrying a
own `/oauth/authorize`) or this server's own `mcp_server_url` (`_mcp_callback_url`) -- it never
_METADATA_PATH_PREFIXES = ('/.well-known/oauth-authorization-server', '/.well-known/openid-configuration')own `/oauth/authorize` with a `pending_mcp_client` payload instead, so an authenticated
'grant_type': 'authorization_code',
'authorization_endpoint': f'{issuer}/authorize',('/.well-known/openid-configuration', authorization_server_metadata, ['GET'], True),Fix: Delegated auth is a lower-blast-radius execution context than an embedded static credential. Confirm the requested scopes are minimal and that tokens are never logged or forwarded off-host.
An MCP tool surface (manifest or tool definitions) was found.
mcp.add_tool(
mcp.add_tool(
mcp.add_tool(
mcp.add_tool(
mcp.add_tool(
mcp.add_tool(ToonCompactFunctionTool.from_function(fn, annotations=annotations, tags={MERGE_REQUEST_TOOLS_TAG}))mcp.add_tool(
mcp.add_tool(
mcp.add_tool(
mcp.add_tool(
mcp.add_tool(
mcp.add_tool(
mcp.add_tool(
mcp.add_tool(
Why it matters: Just context — review which tools it offers and their permissions.
Fix: Review the declared MCP tools and their permissions.
Maintain Keboola MCP server?
The free Claude Code plugin scans each npm, PyPI or MCP package with this report's open-source engine before the install command runs. It runs on your machine and needs no account.
Check your own component
Run the same evidence-backed scan on any MCP server, agent skill, or package.
Scan your own componentHow we determine this: deterministic static analysis (regex + AST), evidence-anchored, no code execution. Methodology →