Skip to content

You're reading this. The skill predicted it. A workflow that turns every post into a calibrated experiment—score, blind-predict, retro, evolve. The future doesn't reward effort, it rewards those who see the pattern first. 1M followers in a month — not luck, system.

Python

chore: CHANGELOG entry for #16 + drop over-broad tests/ gitignore rule

jonc8dd64d548e7
diff --git a/.gitignore b/.gitignore
index d2fdcf9..84801f2 100644
--- a/.gitignore
+++ b/.gitignore
@@ -16,7 +16,6 @@ __pycache__/
.venv/
venv/
.python-version
-tests/
# 用户项目数据(万一开发者在 cheat-on-content 源码目录里跑了 cheat-init)
# 这些文件应该在用户自己的 content project 里,不该污染源码 repo
diff --git a/CHANGELOG.md b/CHANGELOG.md
index efb2d32..ec0a00f 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -8,6 +8,17 @@ All notable changes to cheat-on-content will be documented here.
## [Unreleased]
+### Fixed — douyin-session 运行时路径隐私漏洞(@level5Ninja [#16](https://github.com/XBuilderLAB/cheat-on-content/pull/16))
+
+**问题**:douyin-session adapter 把 `.auth/`(**含抖音登录 cookie**)、debug 截图、report 写进 **skill 源码目录** 而不是用户内容项目——symlink 安装时,用户的会话凭据会落在 cheat-on-content repo 里,有被 commit 的风险。meta-logging hook 还把每条 user prompt 前 120 字存进 `usage.jsonl`,采集过度。
+
+**修复**:
+- 新增 `adapters/perf-data/douyin-session/paths.py` —— 运行时路径 helper(`runtime_project_root` / `auth_dir` / `debug_dir` / `videos_dir`),用 `CHEAT_PROJECT_ROOT` env var + cwd fallback
+- `.auth/` → 用户内容项目根;debug 产物 → `.cheat-cache/douyin-session-debug/`;report/script → 用户项目 `videos/`(不再散落在 skill 源码树)
+- `run.sh` export `CHEAT_PROJECT_ROOT`
+- meta-logging hook 不再存 prompt 摘要——改成只记 `prompt_present`(bool)+ `prompt_chars`(长度)
+- docs(adapter README + state-management.md)同步
+
### Fixed — cheat-shoot DIFF_METRIC 在口语化场景的 v2 误触发(**BREAKING for v2-trigger-logic**)
**问题**:cheat-shoot Phase 3b 用 line-level unified diff 算 `diff_pct = (added + removed) * 100 / orig_lines`。但**创作者真实场景**——draft 是 markdown 长句(一行 ~50 字),拍摄稿是 whisper 转录的口语化短断句(每行 ~5-10 字)——同样的内容会 inflate diff_pct 到 100-200%,触发本不该的 v2 重判。