mantis-meta-agent

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

• Third-party content exposure detected (medium risk: 0.65). The required workflow delegates to multiple subagents (e.g., @mantis_history, @mantis_researcher) that are instructed to analyze the repository and perform deep sweeps, which at runtime can include outsider-authored free text from the target codebase/history (e.g., commit messages, issue-like artifacts, or other repository text) being read and then fed into the LLM context by those subagents.