Install
openclaw skills install @zhan5331/johnzhan-openclaw-skillsProject-based conversation logging for OpenClaw. Use when setting up daily auto-logging across parallel projects, building conversational continuity between sessions, or needing file-backed chat history per project folder. Also covers auto-read of previous day logs on first daily turn.
openclaw skills install @zhan5331/johnzhan-openclaw-skillsAuto-write structured daily conversation summaries into project-specific folders. Each project gets its own directory with daily log files, auto-read on first turn each day, and on-demand manual lookup.
Place this SKILL.md at:
{workspace}/skills/conversation-logger/SKILL.md
# PowerShell
$projects = @("main", "my-project-a", "my-project-b")
$base = "D:\OpenClaw Memory"
foreach ($p in $projects) {
New-Item -ItemType Directory -Path "$base\$p" -Force
}
# Linux / macOS
mkdir -p ~/".openclaw-memory"/{main,my-project-a,my-project-b}
### 🗂️ 项目对话日志(Conversation Logger)
参考 skill: `skills/conversation-logger/SKILL.md`
**自动记录结构:**
- 每次对话结束,**静默**追加结构化摘要到 `D:\OpenClaw Memory\{project}\memory YYYY-MM-DD.md`
- 格式:主题/上下文 → 我的提问 → 决策/结论 → 涉及文件/配置
**每日首次自动调取:**
- 当天第一次对话时,读取前一日对应项目的日志作为上下文
- 当天内如需调取过往日志,手动要求即可
**当前项目文件夹映射:**
| 项目/场景 | 文件夹 |
|---|---|
| 日常主对话 | `main` |
| 项目 Foo | `project-foo` |
| 项目 Bar | `project-bar` |
Each log entry follows this template:
## YYYY-MM-DD HH:mm
**主题/上下文**: [session summary in one line]
**我的提问**: [user's core question]
**决策/结论**: [confirmed decisions, tech choices, config changes]
**涉及文件/配置**: [file paths, config keys, tools used]
Example:
## 2026-07-24 16:42
**主题/上下文**: Setting up Conversation Logger project logging system
**我的提问**: Create per-project conversation log folders with auto-daily recording
**决策/结论**:
- File format: `memory YYYY-MM-DD.md`
- Project folders: `main` / `led-app` / `led-grid` / `slash-fit` / `vired`
- Log granularity: structured summary (topic + question + decision + files)
- Auto-read: previous day's log on first daily turn; manual lookup within same day
- No backfill — start recording from now
**涉及文件/配置**:
- Directories created under `D:\OpenClaw Memory\`
- Skill: `skills/conversation-logger/SKILL.md`
- `AGENTS.md` updated with logging section
| User Request | Agent Action |
|---|---|
| "Show yesterday's logs" | Read previous day's file, display content |
| "Show logs for [date]" | Read specified date's file |
| "Search for [keyword]" | Scan all .md files in the project folder |
| "Show this week's logs" | Iterate last 7 days of files |
Maintain this mapping table in AGENTS.md:
| 项目/场景 | 文件夹 |
|---|---|
| Daily main conversation | `main` |
| Project Alpha | `alpha` |
| Project Beta | `beta` |
When spawning sub-agents, pass the project folder name in the task prompt so the child agent writes to the correct directory.
D:\OpenClaw Memory\ # or ~/.openclaw-memory/ on Unix
├── main\ # Default/main conversations
│ └── memory 2026-07-24.md
├── alpha\ # Project Alpha
│ └── memory 2026-07-24.md
├── beta\ # Project Beta
│ └── memory 2026-07-24.md
└── ...
Cross-platform note: On Linux/macOS, use ~/".openclaw-memory" instead of D:\OpenClaw Memory. Update the base path in AGENTS.md accordingly.
When delegating work to sub-agents via sessions_spawn, add this instruction to the task prompt:
Conversation log: write a structured summary to D:\OpenClaw Memory\{project-folder}\memory YYYY-MM-DD.md after this task.
This ensures sub-agents also contribute to the same daily log.
main folder is the catch-all. Route unknown/unaffiliated conversations there.memory YYYY-MM-DD.md — scripts that parse these should handle spaces.read to check whether today's file already exists before deciding if it's the first turn of the day.