baoyu-electron-extract

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

• Third-party content exposure detected (high risk: 0.85). The skill ingests outsider-authored free text from the target Electron app’s bundled source maps: it reads each runtime-discovered *.js.map file (readFileSync(mapPath, "utf8")) and parses/uses its sourcesContent to write restored source files, which becomes LLM-readable prose if the agent later displays/quotes the restored output.