flux-opencode-rca

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill tells the agent to include the full request ($ARGUMENTS) and to "quote [error messages or stack traces] exactly," which can cause the model to reproduce user-supplied secrets verbatim if they appear in those inputs.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The skill ingests the outsider-authored “Full request: $ARGUMENTS” (user-provided free text) to classify signals and drive the RCA workflow at runtime.