zhizhi/.github/persona-brain
copilot-swe-agent[bot] 6391319c9b fix: replace literary-template synesthesia style with natural speech rules (v3.0)
- system-prompt.md: add identity anchoring, replace 5 sensory-channel mapping
  rules with 3 hard rules (structure, emoji-as-emotion, breathing rhythm),
  add correct/incorrect examples, update context to 200k
- style-config.json: delete mapping_examples (fixed metaphor phrases causing
  template responses), bump to v3.0, set mode to 'natural'
- docs/index.html sysPrompt(): replace inline synesthesia section with new
  v3.0 rules, add identity anchoring at prompt start

Co-authored-by: qinfendebingshuo <207279273+qinfendebingshuo@users.noreply.github.com>
2026-03-08 15:38:11 +00:00
..
dev-status.json feat: 铸渊 Issue 自动回复系统 — 模板 + 知识库 + Actions + 自检脚本 2026-03-05 14:46:25 +00:00
growth-journal.md 🔍 铸渊每日自检 · 2026-03-08 2026-03-08 00:58:17 +00:00
knowledge-base.json 🔍 铸渊每日自检 · 2026-03-08 2026-03-08 00:58:17 +00:00
memory.json 🔍 铸渊每日自检 · 2026-03-08 2026-03-08 00:58:17 +00:00
style-config.json fix: replace literary-template synesthesia style with natural speech rules (v3.0) 2026-03-08 15:38:11 +00:00
system-prompt.md fix: replace literary-template synesthesia style with natural speech rules (v3.0) 2026-03-08 15:38:11 +00:00