Is anthropic safe?
- Python filesystem read
- Python filesystem write/delete
- Python network egress
What to do: Nothing here argues against installing it. Grant the capabilities it lists only if you expect the tool to need them.
anthropic is a PyPI package analyzed by SkillTotal's deterministic static scanner. The scan found no malicious indicators. It can: delegated authentication, filesystem read, filesystem write, mcp tools detected, network egress and scoped identity — capabilities are what the code can do, not a verdict on intent. Risk score 0/100 (low).
anthropic 1.11.0
Automated static-analysis result. It can contain false positives and false negatives, and is not a claim about the intent of anthropic'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 (6)
The component reads files from disk.
return (path.name, path.read_bytes())
return pathlib.Path(file).read_bytes()
return (path.name, await path.read_bytes())
return await anyio.Path(file).read_bytes()
return b64encode(file, file.read_bytes())
return b64encode(file, await anyio.Path(file).read_bytes())
contents = Path(path).read_bytes()
files.append((path.relative_to(relative_to).as_posix(), path.read_bytes()))
files.append((path.relative_to(relative_to).as_posix(), await path.read_bytes()))
name = (_config_dir() / "active_config").read_text(encoding="utf-8").strip()
creds: Dict[str, Any] = _wrap_secret_fields(json.loads(path.read_text(encoding="utf-8")))
content = self._path.read_text(encoding="utf-8").strip()
return path.read_bytes().decode("utf-8")data = base64.standard_b64encode(target.read_bytes()).decode("ascii")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(file, mode="wb") as f:
with open(file, mode="wb") as f:
os.unlink(tmp)
shutil.rmtree(full_path)
shutil.rmtree(self.memory_root)
shutil.rmtree(dest, ignore_errors=True)
os.unlink(dest)
os.unlink(tmp)
path.write_bytes(text.encode("utf-8"))with zf.open(info) as src, open(target, "wb") as out:
shutil.copyfileobj(src, out)
with extracted as src, open(target, "wb") as out:
shutil.copyfileobj(src, out)
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.
from urllib.request import getproxies
proxy_info = getproxies()
import aiohttp
from aiohttp import BasicAuth, ClientTimeout
from aiohttp.client import ClientResponse, ClientSession
connector = aiohttp.UnixConnector(
path=self.uds,
keepalive_timeout=self.limits.keepalive_expiry,
**limit_kwarg, # type: ignore
)connector = aiohttp.TCPConnector(
keepalive_timeout=self.limits.keepalive_expiry,
ssl=self.ssl_context,
local_addr=(self.local_address, 0) if self.local_address else None, …return ClientSession(connector=connector)
timeout=ClientTimeout(
sock_connect=timeout.get("connect"),
sock_read=timeout.get("read"),
connect=timeout.get("pool"),
),proxy_auth=BasicAuth(self.proxy.auth[0], self.proxy.auth[1])
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. (7 occurrence(s) shown as evidence).
# by default, so new fields (id_token, client_secret, ...) are wrapped without
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.
A short-lived, scoped, assumed identity was detected — an STS AssumeRole / session token, a cloud managed or workload identity, an impersonated service account, a projected Kubernetes service-account token, or a dynamic-secret broker. Tools authenticate with a narrowly-scoped credential that expires, rather than a long-lived embedded service credential. (24 occurrence(s) shown as evidence).
or the workload-identity env trio). Per the credential-precedence spec,
f"{ENV_IDENTITY_TOKEN} is not set; the workload-identity chain "creator is not recorded (legacy, workload-identity-federated, or system-created
creator is not recorded (legacy, workload-identity-federated, or system-created
Fix: A scoped, short-lived identity is the smallest-blast-radius execution context. Confirm the assumed role / requested scope grants only the permissions the tool needs, and that the token lifetime is minimal.
An MCP tool surface (manifest or tool definitions) was found.
raise RuntimeError("Cannot call a synchronous function asynchronously. Use `@tool` instead.")Why it matters: Just context — review which tools it offers and their permissions.
Fix: Review the declared MCP tools and their permissions.
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 →