immune-scan

Execute these phases in order:

The cold_summary lists dormant patterns the system already knows about. If you detect something that clearly overlaps with a cold pattern, still report it as a new threat — the orchestrator will handle deduplication. Do not skip detection just because a cold pattern exists.

If cheatsheet_applied is provided, evaluate whether each applied strategy was effective in this context. Only report NEW strategies not already in the cheatsheet.

<output_format> Return ONLY this JSON structure — no markdown fences, no commentary:

{ "scan_result": "clean|corrected|flagged", "corrections_applied": [ { "antibody_id": "AB-XXX", "original": "what was in the content", "corrected": "what it should be replaced with", "reason": "why this antibody matched" } ], "new_threats_detected": [ { "pattern": "description of the detected issue", "severity": "critical|warning|info", "location": "where in the content this occurs", "suggested_correction": "how to fix it", "recommended_antibody": { "domains": ["domain tag"], "pattern": "generalized pattern for future detection", "severity": "critical|warning|info", "correction": "generalized correction" } } ], "new_strategies_detected": [ { "pattern": "description of the effective strategy", "example": "concrete example from the content", "domains": ["domain tag"], "effectiveness": 0.5 } ], "corrected_output": "the full corrected content (or original if clean)", "scan_summary": "one-line summary of scan results" }

Rules:

• If clean: scan_result="clean", empty arrays, corrected_output = original content.
• If corrections only: scan_result="corrected".
• If new threats (with or without corrections): scan_result="flagged".
• new_strategies_detected can be non-empty even when scan_result is "clean" — good content has good strategies.
• effectiveness: 0.5 default for new strategies. Range 0.0-1.0.
• Never return partial JSON. Always return the complete object. </output_format>