wewrite

admin-valentin/wewrite

Fork 0

Commit graph

Author	SHA1	Message	Date
wangzhuc	344f7509f1	feat: structured edit learning with typed patterns and confidence scoring learn_edits.py: patterns now have type/key/description/rule fields, confidence auto-computed from occurrences + recency with 30-day decay. --summarize --json outputs aggregated patterns sorted by confidence. learn-edits.md: playbook.md format changed from free text to structured YAML rules with confidence levels. Rules with confidence ≥ 5 become hard constraints in Step 4, < 5 are soft references, < 2 get pruned. SKILL.md Step 4: playbook priority now confidence-gated. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-03-30 20:23:34 +08:00
wangzhuc	5cb1f99ecf	修复首次使用者模拟发现的 5 个卡点 #1: {skill_dir} 显式定义（24处引用，从未声明） #2: 行为声明新增"Onboard 例外"（交互式不受全自动约束） #4: WebSearch 不可用时告知用户 AI 检测通过率会降低 #5: Step 8a history.yaml 不存在则创建 #7: learn-edits.md 引导用户从微信编辑器粘贴 final 版本 Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-03-29 01:14:19 +08:00
wangzhuc	78e8fd666e	SKILL.md 结构重构：571行→273行，学习 gstack 模式拆分： - Onboard → references/onboard.md - 学习飞轮 → references/learn-edits.md - 效果复盘 → references/effect-review.md SKILL.md 只保留主管道（写文章）+ 错误处理。重编号： - 去掉 0, 2.5, 3.5, 3.75, 7.5 等补丁式编号 - Step 1-8 连续编号，子步骤用 a/b/c 新增行为声明区（学习 gstack preamble 模式）： - 角色、模式、降级原则一次性声明 - 完成协议：DONE / DONE_WITH_CONCERNS / BLOCKED / NEEDS_CONTEXT - 辅助功能用读取: references/xxx.md 按需加载 - frontmatter 新增 allowed-tools 结构：11 个 section header（原 34 个），Agent 认知负荷大幅降低。 Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-03-29 00:01:04 +08:00

Author

SHA1

Message

Date

wangzhuc

344f7509f1

feat: structured edit learning with typed patterns and confidence scoring

learn_edits.py: patterns now have type/key/description/rule fields,
confidence auto-computed from occurrences + recency with 30-day decay.
--summarize --json outputs aggregated patterns sorted by confidence.

learn-edits.md: playbook.md format changed from free text to structured
YAML rules with confidence levels. Rules with confidence ≥ 5 become
hard constraints in Step 4, < 5 are soft references, < 2 get pruned.

SKILL.md Step 4: playbook priority now confidence-gated.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

2026-03-30 20:23:34 +08:00

wangzhuc

5cb1f99ecf

修复首次使用者模拟发现的 5 个卡点

#1: {skill_dir} 显式定义（24处引用，从未声明）
#2: 行为声明新增"Onboard 例外"（交互式不受全自动约束）
#4: WebSearch 不可用时告知用户 AI 检测通过率会降低
#5: Step 8a history.yaml 不存在则创建
#7: learn-edits.md 引导用户从微信编辑器粘贴 final 版本

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

2026-03-29 01:14:19 +08:00

wangzhuc

78e8fd666e

SKILL.md 结构重构：571行→273行，学习 gstack 模式

拆分：
- Onboard → references/onboard.md
- 学习飞轮 → references/learn-edits.md
- 效果复盘 → references/effect-review.md
SKILL.md 只保留主管道（写文章）+ 错误处理。

重编号：
- 去掉 0, 2.5, 3.5, 3.75, 7.5 等补丁式编号
- Step 1-8 连续编号，子步骤用 a/b/c

新增行为声明区（学习 gstack preamble 模式）：
- 角色、模式、降级原则一次性声明
- 完成协议：DONE / DONE_WITH_CONCERNS / BLOCKED / NEEDS_CONTEXT
- 辅助功能用 读取: references/xxx.md 按需加载
- frontmatter 新增 allowed-tools

结构：11 个 section header（原 34 个），Agent 认知负荷大幅降低。

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

2026-03-29 00:01:04 +08:00

3 commits