removed old input files, adding chapters 7-8 and

world/magic/exploits-log.md

@@ -20,3 +20,12 @@ Every exploit Phelan uses, logged for continuity tracking.
 - **Method:** Generated ~400 simultaneous low-level magical inputs across all monitored channels. Random noise — no pattern to match, no precision to counter. Response array saturated trying to process contradictory data from 400 sources. Detection still ran but response mechanism locked up. Leon walked through.
 - **Item recovered:** Mallory focusing crystal, pre-Compact, original inscription. Sold for 1,200 silvers to anonymous buyer.
 - **Notes:** Brute-force cousin to Phelan's precision approach. Same philosophical family — exploit the working's own design assumptions. Important concept: input flooding as a technique.
+
+## Exploit #3: Death Ward — Signature Forgery & Self-Redirect
+- **Chapter:** Ch08
+- **Target:** Death ward on second floor, third door, Greymarch Barrows. Pre-Compact construction. Converter-type ward — routes all external magical input through 26 conversion nodes (three concentric rings) and outputs as thermal energy. Designed by an engineer, not a theorist. Built to handle brute-force attacks (Leon's 400-input flood converted to heat without effect).
+- **Flaw:** Imperfect conversion ratio (~2% leakage). The conversion process cannot be 100% efficient — fractionally narrower output channels than input channels. The 2% unconverted energy circulates internally through the routing matrix as the ward's own resting signature. This internal traffic is invisible to the external-facing conversion nodes, which are calibrated outward. The leakage broadcasts the ward's internal energy profile continuously — frequency, amplitude, phase offsets at seven routing junctions.
+- **Method:** Three-phase exploit. (1) **Reconnaissance:** Deep Flaw Sight analysis mapped the ward's full architecture and the internal circulation signature over 8+ observation cycles. Passive, low energy cost, time-intensive. (2) **Forgery & Injection:** Phelan matched his magical output to the ward's internal signature — frequency (12.7 cycles/standard interval), amplitude, and phase offsets at all seven junctions. Fed forged energy through the 2% leakage seam into the internal channels. Ward accepted it as its own traffic. Built to ~12% of internal circulation volume. Sustained concentration required — any deviation in phase offset would trigger conversion nodes. High energy cost (reserves dropped from ~80% to ~50%). (3) **Redirect:** Altered routing destination on forged energy at junction five. Ward's conversion nodes processed the altered traffic as internal, converting the ward's own routing architecture to heat. Self-sustaining cascade — converted architecture produced debris that circulated to more conversion nodes, which converted more architecture. Ward consumed itself.
+- **Cost:** Extreme. Reserves from ~80% to ~15%. Sustained precision output for extended duration. Physical symptoms: hand tremor, sweating, neck wound aggravated. Flaw Sight flickering at edges post-exploit. Crash incoming but not yet arrived at chapter's end.
+- **Flaw Sight used:** Yes — full deep analysis engagement. Extended duration. The reconnaissance phase required the longest continuous Flaw Sight use to date.
+- **Notes:** Most sophisticated exploit to date. Required theory developed in ch06 (conversation with Leon about conversion ratio and leakage). The exploit doesn't overpower the ward — it convinces the ward to destroy itself using its own conversion function. Philosophical principle: if a system has an imperfect process, the imperfection IS the access point. The ward's strength (efficient conversion) was also its vulnerability (the conversion couldn't distinguish self from forged self). Leon's brute-force attempt (Exploit #2 context) established that direct assault was futile, which forced the analytical approach.