openclaw-codex-deep-search

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

• Third-party content exposure detected (high risk: 0.90). The skill's scripts/search.sh SEARCH_INSTRUCTION explicitly directs the Codex CLI to "Search the web for the following query" and incrementally write findings (including source URLs) to data/codex-search-results/.md which the agent reads and summarizes, meaning it fetches and ingests untrusted public web content as part of its workflow and could be subject to indirect prompt injection.