escalation-l2

What problem does it solve?

L2 escalation automates the autonomous proposal of modest architectural extensions when L1 yields NO-VIABLE-PROPOSAL, writing a formal appendix entry to ARCHITECTURE.md and preserving the architecture lock_hash.

Core Features & Use Cases

• Three parallel subagents (New-Function-in-Existing-Module, New-Module-in-Existing-Layer, Pre-Listed-Library) run to generate balanced, low-risk proposals.
• Autonomous, no human gate: the skill surfaces an appendix entry and optionally triggers L3 if all lenses fail.
• Writes an additive appendix entry to ARCHITECTURE.md, preserving the lock_hash and enabling future audits.
• Works after L1 exhausts its lenses, when operator triggers escalation or L1 fallback flag is used.

Quick Start

Use the escalation-l2 skill after L1 exhausts its lenses to autonomously propose modest architecture extensions and append an appendix entry.