enterprise-search

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

• Third-party content exposure detected (high risk: 0.95). Outsider-authored free text is ingested when the workflow calls /rest/api/v1/getdocuments via scripts/es_read.sh, which returns content.fullTextList (full document text from indexed sources like Slack/Confluence/Jira) and prints it into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.90). The scripts call the configured GLEAN_BASE_URL at runtime (e.g. https://your-company-be.glean.com/rest/api/v1/getdocuments and /rest/api/v1/search) to fetch full document text that is then emitted for use as agent context, so external content returned by those URLs can directly control prompts or instruct the agent.