compose‑lint, a Docker Compose linter, could print untrusted string data that contains GitHub Actions or Azure Pipelines workflow‑command syntax. An attacker who controls a compose file can inject lines such as legacy ##[ commands or double‑colon prefixed commands. When the linter runs in a workflow, the runner interprets these injected commands and can create false annotations, hide or mask log sections, or trigger other runner actions, thereby altering the CI output seen by reviewers. The vulnerability is reflected in CWE‑117 (Insecure Output Encoding) and CWE‑150 (Information Exposure through Logs).

The problem exists in all releases of tmatens/compose-lint before version 0.31.0. Any project using the compose‑lint action with a pre‑0.31.0 binary is affected, regardless of the environment in which the action is executed.

The vulnerability has a moderate CVSS score of 5.3 and no EPSS score is available; it is not listed in the CISA KEV catalog. The likely attack vector is a malicious contributor submitting a pull request that includes a compose file with crafted workflow‑command syntax. While the vulnerability does not provide a direct code execution path, it can mislead reviewers and conceal real CI failures by manipulating annotations and log visibility, thereby reducing the reliability of the review process and potentially allowing further malicious activity.