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.
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_taskwith 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, orcosmetic - 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
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.