herdr-pm-agent

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

• Third-party content exposure detected (high risk: 0.78). The required runtime workflow reads the managed agent’s pane scrollback via pm.py read --source recent-unwrapped / pm.py diagnose (which calls pane read), and that pane content is outsider-authored free text produced by the managed agent (not the operating user), so it can include indirect prompt-injection text that gets fed back into the conductor’s LLM context.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)