jira-read

SUSPICIOUS. The skill’s purpose broadly matches its Jira-reading capabilities, and its data flows target Jira rather than an unrelated third-party proxy. However, it requires raw credential-file access, forwards PAT/cert material to unseen local scripts and optional MCP tooling, and explicitly permits disabling TLS verification, which creates a meaningful credential-interception and trust risk disproportionate to a simple read helper.

The code is not evidently malware, but it contains a significant credential-leak vulnerability: unrestricted redirects can cause Bearer or Basic credentials and mutual-TLS authentication to be sent to an arbitrary host. TLS verification can also be disabled through JIRA_SSL_VERIFY=false, weakening transport security. Redirects should be restricted to the configured Jira origin, and TLS verification should not be disabled except under tightly controlled development conditions. The output path should be treated as trusted because it permits arbitrary file writes by the invoking user.