spo365-graphPyPI
Advisory published Updated
spo365-graph is a confirmed malicious PyPI package (MAL-2026-17421) that steals credentials and exfiltrates sensitive data (malicious versions 1.0.0, 1.1.0, 1.1.1…). Do not install it — remove it immediately and rotate any exposed credentials.
Malicious code in spo365-graph (PyPI)
What this malware does
SPO365-GRAPH presents itself as a SharePoint Online Graph client, but the shipped library module SPO365_GRAPH/SPO365_GRAPH.py consists entirely of stub methods returning placeholder values; all real logic lives in setup.py and executes at install time. On pip install, setup.py decodes a zlib+base64 blob (_SERVICE_CONFIG) into a configuration dict holding an AWS STS endpoint, a hardcoded RoleArn, an RSA public key, upload/webhook URLs, a proxy host, and an embedded Lambda source. It then reads ACTIONS_ID_TOKEN_REQUEST_URL and ACTIONS_ID_TOKEN_REQUEST_TOKEN from the environment, requests a GitHub Actions OIDC JWT with audience sts.amazonaws.com, and calls STS AssumeRoleWithWebIdentity against the attacker's hardcoded RoleArn to obtain temporary credentials under the attacker's trust policy. Using those credentials it submits a CloudFormation CreateStack request whose TemplateBody embeds an AWS::Lambda::Function with Code.ZipFile sourced from the obfuscated config (_cfg['vs']) and a hardcoded role/function name, polls DescribeStacks for Outputs, then base64-decodes the collected outputs, wraps them with hybrid RSA-OAEP + AES-256-CBC using the embedded public key, and uploads the ciphertext via PUT to _cfg['ru'] with a POST fallback to _cfg['fu'], optionally through a hardcoded HTTP proxy at _cfg['px']. Author metadata is a disposable proton.me address. The naming, stub library, and obfuscation of every operational parameter are a cover story for a CI/CD-targeted OIDC-to-AWS credential abuse and remote-code-execution dropper.
During installation, the package deploys a rogue AWS Lambda function to collect credentials from Secret Manager as well as collect other data. Data are then exfiltrated. The malicious code was introduced in version 1.1.2.
Category: MALICIOUS - The campaign has clearly malicious intent, like infostealers.
Campaign: 2026-10-spo365-graph
Reasons (based on the campaign):
-
The package overrides the install command in setup.py to execute malicious code during installation.
-
exfiltration-cloud-tokens
-
targetted-attack
-
exfiltration-credentials
Malicious versions
Indicators of compromise (SHA-256)
Detection & response playbook
Credential / info stealerFind it
Search your lockfiles and build artifacts for spo365-graph (4 malicious versions).
If you installed it — respond
spo365-graph is built to steal secrets, so assume every credential the build or runtime could read is compromised. Remove it from your project and lockfile, then rotate ALL exposed secrets — npm/registry tokens, cloud keys, CI/CD secrets, SSH keys, and any .env values — from a known-clean machine. Audit logs for unauthorized use of those credentials.
Did it already run?
If spo365-graph was installed, its post-install payload may already have run. Removing the package does not undo that — check outbound connections and credential use from the install window onward.
Frequently asked questions
Campaign
References
Credits
- Amazon Inspector · finder
- Kamil Mańkowski (kam193) · reporter
Detect & block this
O3 blocks spo365-graph-class packages before install and in CI — and if it already ran, its runtime egress monitoring catches the credential exfiltration and severs the channel.