Claude Session Logger

Claude Session Logger

A local MCP server that records completed tasks to daily JSONL files and promotes substantial work to a cumulative weekly Markdown worklog, providing persistent, searchable logs of AI-assisted productivity.

Category
访问服务器

README

Claude Session Logger

A local MCP server that keeps a running log of what you accomplish across Claude sessions. Each task you complete gets recorded to a daily JSONL file, and substantial work is automatically promoted to a cumulative weekly worklog — giving you a persistent, searchable record of your AI-assisted productivity.

Why

Claude conversations are ephemeral. You finish a session, close the window, and the work you did fades into chat history. This server solves that by giving Claude a set of tools to log tasks as they happen. At the end of a week you have a clean Markdown summary of the important things you shipped, without the noise of minor fixes and formatting tweaks.

Features

  • Daily task logs — JSONL files with one entry per completed task, including summary, detail notes, tags, project name, status, and estimated duration.
  • Weekly worklog — A cumulative Markdown file (Monday–Sunday) that automatically captures substantial work. Tasks with detail notes, 15+ minute duration, or meaningful tags are included; trivial changes like typos and lint fixes are filtered out.
  • UTF-8 text sanitization — All text fields are cleaned on input: NFKC Unicode normalization, control character removal, surrogate stripping, and whitespace collapsing. No broken characters in your logs.
  • Rebuild from source — Weekly worklogs can be regenerated from raw daily data at any time, so edits to daily logs are always reflected.

Requirements

  • Python 3.11+
  • Claude Desktop or Claude Code

Installation

git clone https://github.com/YOUR_USERNAME/claude-session-logger.git
cd claude-session-logger
pip install -e .

Verify:

python -c "from session_logger.server import mcp; print('OK')"

Setup

Claude Desktop (Windows/macOS)

Open the config file:

# Windows
notepad %APPDATA%\Claude\claude_desktop_config.json

# macOS
open ~/Library/Application\ Support/Claude/claude_desktop_config.json

Add the server to the mcpServers section:

{
  "mcpServers": {
    "session-logger": {
      "command": "python",
      "args": ["-m", "session_logger.server"],
      "cwd": "C:\\path\\to\\claude-session-logger"
    }
  }
}

Restart Claude Desktop after saving.

Claude Code (CLI)

claude mcp add session-logger --scope user -- python -m session_logger.server

To enable automatic logging, add this to your CLAUDE.md or project custom instructions:

After completing each distinct task, call session_log_task with a one-line summary, relevant tags, and status. Log automatically whenever you commit changes with Git.

Tools

Tool Description
session_log_task Record a task to today's daily log. Substantial entries are auto-appended to the weekly worklog.
session_view_log View a day's entries in Markdown or JSON format. Defaults to today.
session_list_days List all dates with log entries, reverse chronological.
session_daily_summary Aggregate stats for a day: task counts by status, total duration, tags, projects.
session_weekly_worklog View the cumulative Mon–Sun worklog. Use rebuild=true to regenerate from raw daily data.

Task fields

Field Required Description
summary Yes One-line description of the completed work
detail No Longer notes, context, or follow-up items
status No completed (default), partial, blocked, or deferred
tags No Category labels like bugfix, refactor, docs
project No Project or repo name for grouping
duration_minutes No Estimated time spent (0–1440)

How the weekly worklog works

When a task is logged, the server classifies it as substantial or minor:

Substantial (included in the weekly worklog):

  • Has detail text
  • Duration of 15 minutes or more
  • Carries tags beyond trivial categories

Minor (daily log only):

  • Tagged only with typo, formatting, whitespace, lint, cleanup, trivial, style, nit, or cosmetic
  • No detail text and under 15 minutes

Substantial tasks are appended to worklog-YYYY-WNN.md as they're logged. Each ISO week (Monday–Sunday) gets its own file. The session_weekly_worklog tool can also rebuild the worklog from raw daily JSONL data if entries are edited or deleted after the fact.

File structure

~/.claude-session-logs/
├── 2026-03-31.jsonl            # Daily task log (one JSON object per line)
├── 2026-04-01.jsonl
├── worklog-2026-W14.md         # Weekly worklog (Mon Mar 30 – Sun Apr 5)
└── ...

Configuration

Set SESSION_LOG_DIR to change the log directory:

# Linux/macOS
export SESSION_LOG_DIR=~/Documents/claude-logs

# Windows (persistent)
setx SESSION_LOG_DIR "C:\Users\you\Documents\claude-logs"

Default: ~/.claude-session-logs/

Reading logs outside Claude

# View today's raw entries
cat ~/.claude-session-logs/2026-03-31.jsonl | python -m json.tool --json-lines

# View this week's worklog
cat ~/.claude-session-logs/worklog-2026-W14.md

# Count tasks by project across all logs
cat ~/.claude-session-logs/*.jsonl | python -c "
import sys, json
from collections import Counter
c = Counter(json.loads(l).get('project','(none)') for l in sys.stdin)
for proj, n in c.most_common(): print(f'{n:4d}  {proj}')
"

Project layout

claude-session-logger/
├── pyproject.toml
├── README.md
└── session_logger/
    ├── __init__.py
    ├── sanitize.py     # UTF-8 text cleaning and tag normalization
    ├── models.py       # Pydantic schemas: TaskEntry, DailySummary, WeeklyWorklogEntry
    ├── store.py        # File I/O: daily JSONL, weekly Markdown, summaries
    └── server.py       # FastMCP server and tool definitions

License

MIT

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选