Roo Activity Logger

Roo Activity Logger is an MCP server that records AI coding assistant actions into structured, searchable JSON files. Designed for software developers working with assistants like Roo Code, Cline, and Claude Code, it captures developmental activity including command executions, code generation, file updates, error events, conversations, and architectural decisions. By cataloging developer actions alongside explicit intention, operational context, hierarchy relationships, and execution status, the server allows coding agents to preserve institutional context across sessions. Instead of losing state when context windows clear, developers and models can query historical records through dedicated search tools. The server operates locally on your filesystem, saving log files to user-specified directories without relying on external cloud databases. This structured traceability helps teams debug agent errors, inspect automated code modifications, reconstruct development narratives, and safely resume interrupted tasks with full awareness of prior decisions.

Category: Developer Tools & Code Intelligence

Tags: activity-tracking, coding-assistant, json, logging

Visit Roo Activity Logger

How to install and configure Roo Activity Logger

1. Open your MCP client configuration file, such as your Cline, Roo-Code, or Claude Code settings. 2. Add the roo-activity-logger configuration under mcpServers: json { "mcpServers": { "roo-activity-logger": { "command": "npx", "args": ["-y", "github:annenpolka/roo-logger"], "env": {}, "disabled": false } } } 3. Alternatively, install locally by cloning https://github.com/annenpolka/roo-logger.git, running npm install, and running npm run build, then point command to node and args to /path/to/your/local/roo-logger/dist/index.js. 4. Configure prompts in rule files or CLAUDE.md instructing the assistant to call log_activity and search_logs during tasks.

What you can do with Roo Activity Logger

• Restoring previous task context across agent sessions by searching historical activity entries using date ranges, keywords, or activity types. - Auditing automated file modifications and command executions by reviewing structured JSON summaries with explicit intention and context fields. - Tracking error occurrences and debugging failed shell commands encountered by AI assistants during automated refactoring workflows. - Establishing parent-child execution hierarchies to track complex multi-step coding objectives and their corresponding subtask sequences. - Inspecting decisions made by autonomous coding agents before allowing further modifications to project files or architectural dependencies.

Key facts

• https://github.com/annenpolka/roo-logger
• Developer Tools & Code Intelligence, Files, Documents & PDFs
• activity-tracking, coding-assistant, json, logging

Part of MCP Servers

What is Roo Activity Logger?

Roo Activity Logger is an open-source Model Context Protocol server that captures actions executed by AI coding assistants. It writes operations like code generation, command executions, and file edits into structured JSON records on your local drive for context retrieval.

Which AI clients work with Roo Activity Logger?

It is built for MCP-compatible AI development environments and coding assistants, including Roo Code, Cline, and Claude Code. Any client supporting standard Model Context Protocol tool execution can interface with the server.

What tools does Roo Activity Logger provide?

The server exposes three primary tools: log_activity to record actions with context and metadata, get_log_files to list available JSON activity files in a directory, and search_logs to query entries by date, text, activity type, or hierarchy.

What license does Roo Activity Logger use?

Roo Activity Logger is released as open-source software under the MIT license, permitting free commercial and personal use, modification, and distribution.

Does Roo Activity Logger require a cloud database?

No. The server writes records directly to local JSON files using an absolute directory path specified in the tool arguments. It creates the destination directories automatically if they do not exist.