write-testrail-cases

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

• Third-party content exposure detected (high risk: 0.90). The required workflow feeds the outsider-authored Jira ticket free-text (story Description/Acceptance Criteria) into the agent’s LLM context via step 3 (“Read the User Story / Jira ticket”), using Jira MCP when an ID is given.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly fetches GitHub source files at runtime (used to enrich the model context), e.g. the repository reference "folio-org/ui-agreements" is listed and used to pull translations/package.json fragments that are injected into the agent's in-memory context—this is a runtime fetch of external content that can directly influence prompts.