context-budget-audit

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/audit_context_budget.py, the runtime workflow scans local evidence files under default_evidence_roots() (e.g., ~/.claude/history.jsonl, ~/.codex/history.jsonl, ~/.pi/agent/run-history.jsonl) and reads their contents via read_evidence()/iter_evidence_files(), so any outsider-authored text present in those transcripts can be ingested for regex matching.