Skip to main content
Glama

worklog

CI Release Image License: MIT

Your AI agents keep your work log. worklog is an MCP server: Claude Code, Claude Desktop, claude.ai and other MCP clients record what they finished – one entry per result – and you get it back as a searchable log, reports and a timesheet.

Features

  • For agents: tools to log, correct and search entries, and prompts for weekly and monthly reports, a stand-up and for logging the history of an old project from its git log afterwards.

  • Web view: entries with full text search and filters, overview with activity calendar, timesheet with rounding, project notes, duplicate detection, entries by hand, light and dark, English and German. More

  • Your data: export as Markdown, CSV or JSON; JSON import between instances; import of earlier Claude Code, claude.ai and JetBrains Junie sessions. More

  • Reports: weekly or monthly report as a file and as a notification to ntfy, Discord, Telegram, e-mail and more (Apprise).

  • Safe by default: one token per agent (stored as a hash, the agent name comes from the token), several users strictly separated, optional sign-in through your reverse proxy (e.g. Authelia).

  • Small: one container, SQLite, HTTP (Streamable HTTP) or stdio; images for amd64 and arm64.

Related MCP server: joa

Quick start

Run the server on your machine and connect Claude Code to it. For a server that other devices or claude.ai reach over HTTPS, continue with the setup guide.

With Docker (image for amd64 and arm64 from GitHub Container Registry):

mkdir -p ~/worklog-data && sudo chown 99:100 ~/worklog-data      # container runs as UID 99; skip on Docker Desktop
docker run -d --name worklog-mcp --restart unless-stopped \
  -p 127.0.0.1:8000:8000 -v ~/worklog-data:/data ghcr.io/axute/worklog-mcp:latest

docker exec worklog-mcp worklog token create laptop --owner me   # prints the token (wl_...) once

Without Docker (Python 3.12 or newer):

python3 -m venv ~/.worklog && . ~/.worklog/bin/activate
pip install git+https://github.com/axute/worklog-mcp.git           # needs git

export WORKLOG_DB=~/worklog/worklog.db WORKLOG_HOST=127.0.0.1    # default: /data/worklog.db on 0.0.0.0
worklog-server                                                    # runs in the foreground; Ctrl+C stops it

# in a second terminal, with the same WORKLOG_DB:
worklog token create laptop --owner me                            # prints the token (wl_...) once

Connect Claude Code and check:

claude mcp add --transport http --scope user worklog http://127.0.0.1:8000/mcp \
  --header "Authorization: Bearer wl_..."
claude mcp list                                                   # worklog … ✓ Connected

In a new Claude Code session, ask for example "Log to worklog: project test, summary 'worklog is running'". The entry shows up in the web view at http://127.0.0.1:8000/ui/ (sign in there with a token that has the read scope; laptop has it), or with worklog export --owner me. To make agents log on their own, add the agent rule to ~/.claude/CLAUDE.md.

Documentation

Development

./scripts/test.sh          # pytest in a python:3.12-slim container, dependencies cached in a Docker volume
./scripts/test.sh -k auth  # extra pytest arguments are passed through

The tests start the real server on 127.0.0.1 with a temporary database and talk to it with the FastMCP client; no outside network is used.

License

MIT, see LICENSE.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      D
      maintenance
      Enables AI agents to track, search, and retrieve their progress across projects with persistent memory using SQLite storage and LLM-powered summarization. Supports logging completed work, searching previous entries, and retrieving context for multi-step or multi-agent workflows.
      26 npm
      MIT
    • A
      license
      Not graded
      quality
      D
      maintenance
      Persistent activity journal for AI agents - enables logging and querying decisions, changes, errors, and observations across sessions.
      3 npm
      1
      MIT
    • A
      license
      Not graded
      quality
      C
      maintenance
      Captures AI agent runs and turns them into tamper-evident execution records showing tool use, timing, failures, recoveries, and human interventions. Records can be inspected, exported, and verified offline.
      MIT