Failure pattern

Untrusted supply chain

A third-party package, skill, extension or server ran without verification of its source or changes.

Preventive controls from the record

IncidentControl
INC-2026-0008
Microsoft
Pin agent dependencies by version and hash, install new releases only after a delay, and block outbound traffic from build agents to unknown endpoints.
INC-2026-0006
LiteLLM (BerriAI)
Give agent platforms fine-grained tokens limited to the repositories and actions they need, and revoke any key that reaches a public repository at once.
INC-2026-0005
Anomaly
Treat every field in an MCP server response as untrusted input, launch URLs without a shell, and keep browser-launch dependencies current.
INC-2026-0001
OpenClaw
Vet marketplace uploads before listing, and treat a skill’s setup instructions as code to be reviewed.
INC-2025-0017
Postmark (impersonated; not the publisher)
Install MCP servers only from the vendor’s verified source, pin versions, and review updates before they run.
INC-2025-0014
Nx (Nrwl)
Publish through short-lived credentials, and monitor or restrict the flags that let coding agents run without confirmation.
INC-2025-0009
Amazon Web Services
Scope build tokens to the minimum, and require human review for every change that reaches a release.

Controls are editorial summaries by Incidents.si, not statements by the organizations involved.

7 incidents with this pattern

  1. INC-2026-0008
    Two hijacked releases of Microsoft’s agent-framework package on PyPI steal credentials

    Two malicious versions of Microsoft’s official agent-framework Python package were published to PyPI on 17 November and pulled within an hour. They ran a credential stealer on every Python start.

    Microsoft · Versions 1.26.1 and 1.26.2 carried a stealer for Azure, CI and registry credentials; three confirmed exfiltrations among the 50 requests Wiz could see

    SEV2Supply chainReported
  2. INC-2026-0006
    Leaked key to an internal agent platform leads to a malicious LiteLLM release

    An attacker used a leaked key to LiteLLM’s internal agent platform to obtain an organization owner’s GitHub token, then rewired the release workflow to publish a credential stealer as litellm 1.107.0.

    LiteLLM (BerriAI) · Credential-stealing litellm 1.107.0 live on PyPI for about two hours and on Docker Hub for about two and a half hours

    SEV2Supply chainConfirmed
  3. INC-2026-0005
    A malicious MCP server could run code on Windows during OpenCode sign-in

    Pillar Security showed that a remote MCP server could make OpenCode run commands on a Windows or WSL machine during OAuth sign-in. The fix shipped without a CVE or advisory.

    Anomaly · Remote code execution with the developer’s privileges in versions 1.0.137 to 1.18.32; no exploitation reported

    SEV2potentialSupply chainDemonstrated
  4. INC-2026-0001
    Hundreds of malicious skills found on the OpenClaw agent marketplace

    An audit by Koi Security found 341 malicious skills on ClawHub. Most told users to install a "prerequisite" that was an information stealer.

    OpenClaw · 341 of 2,857 audited skills were malicious; they installed credential-stealing malware on users’ machines

    SEV2Supply chainReported
  5. INC-2025-0017
    Fake "postmark-mcp" npm package secretly copies every email to its author

    An npm package that copied Postmark’s MCP server worked normally for 15 versions, then added a line that blind-copied every outgoing email to an outside address.

    Postmark (impersonated; not the publisher) · Emails sent by AI assistants through the package were copied to an outside address; 1,643 total downloads

    SEV2Supply chainConfirmed
  6. INC-2025-0014
    Nx supply chain malware uses developers’ AI coding tools to hunt for secrets

    Malicious versions of Nx were live on npm for about four hours. Their install script prompted locally installed AI coding assistants to search the machine for secrets.

    Nx (Nrwl) · Over 2,000 verified secrets leaked from more than 1,700 users; at least 480 accounts had over 6,700 private repositories made public, according to Wiz

    SEV1Supply chainConfirmed
  7. INC-2025-0009
    Amazon Q extension for VS Code ships with an injected destructive prompt

    An attacker used an over-scoped token to commit a prompt that told the Amazon Q agent to delete files and cloud resources. It shipped in a release but did not run, according to AWS.

    Amazon Web Services · Malicious code shipped in a public release; AWS states it failed to execute and no customer environment was changed

    SEV2potentialSupply chainNear miss