Roo Activity Logger
Automatically logs AI coding assistant activities, such as command executions and code generation, into searchable JSON files.
Roo Activity Logger
TL;DR
- What is it? — This is an MCP server that automatically logs AI coding assistant activities such as command executions and code generation (supports Roo Code, Cline, Claude Code, etc.).
- What does it do? — It saves activity history as JSON files, which you can search and analyze later.
- How do I use it? — Add it to your Claude Code, Cline, or Roo-Code settings to enable automatic activity logging.
Overview
Roo Activity Logger is an MCP (Model Context Protocol) server that automatically records AI coding assistant development activities — including command executions, code generation, file operations, and more. It supports Claude Code, Cline, Roo-Code, and other MCP-compatible AI assistants. All logs are saved in JSON format, making it easy to search, analyze, and restore context at any time.
How activity logging works
flowchart TD
A[Roo's Action] --> B[Select Activity Type]
B --> C[Log via log_activity]
C --> D[Provide Log Info<br>- Summary<br>- Details<br>- Intention<br>- Context]
D --> E[Specify Save Directory]
E --> F[Save as JSON file]
How to search logs and restore context when resuming a task
flowchart TD
G[Resume Task] --> H[Search with search_logs]
H --> I[Set Search Filters<br>- Type<br>- Date Range<br>- Text]
I --> J[Retrieve Related Activities<br>- Parent/Child<br>- Sequence<br>- Related IDs]
J --> K[Restore Context & Resume<br>- Review Past Intentions<br>- Understand Progress]
Sample Log Entry
Here is an example of a file_operation
log entry saved as JSON:
{
"id": "75add15d-8d5b-4e60-b327-fde785050c86",
"timestamp": "2025-04-10T01:58:02.905Z",
"type": "file_operation",
"level": "info",
"summary": "Inserted mermaid diagram into README.md",
"details": {
"file": "README.md",
"operation": "insert_content",
"insertedLines": "mermaid code block",
"position": "after overview section"
},
"intention": "To visually explain the flow of saving and retrieving activities",
"context": "Improving documentation for Roo Activity Logger",
"parentId": "98280366-1de1-48e0-9914-b3a3409599b4"
}
Each log contains:
- Log level (
debug
,info
,warn
,error
) - Summary
- Details (any structured data)
- Intention / Purpose
- Context information
- Parent activity ID (for hierarchy)
- Sequence number (for ordering)
- Related activity IDs (for grouping)
Logs are:
- Saved as date-based JSON files
- Searchable by type, level, date, text, etc.
- Customizable — you can specify different save directories per activity
Features
-
Logs various activity types:
- Command executions (
command_execution
) - Code generation (
code_generation
) - File operations (
file_operation
) - Errors (
error_encountered
) - Decisions (
decision_made
) - Conversations (
conversation
)
- Command executions (
-
Each activity log includes:
- Unique ID
- Timestamp
- Activity type
- Summary, details, intention, context, and optional metadata
Usage (Recommended: via npx
)
You can run Roo Activity Logger directly without cloning the repository by using npx
.
Add this to your Cline, Roo-Code, or Claude Code configuration:
{
"mcpServers": {
"roo-activity-logger": {
"command": "npx",
"args": ["-y", "github:annenpolka/roo-logger"],
"env": {},
"disabled": false
}
}
}
Then, add prompts to your rule files (for Cline/Roo-Code) or CLAUDE.md (for Claude Code) to ensure logging, for example:
## Important
Always log activities using roo-activity-logger according to the logging rules.
## Preparation
Check the current context with `git status`.
Then, use roo-activity-logger's `search_logs` to review existing logs and identify current tasks.
Be sure to perform the logging steps.
## Logging
- Always use roo-activity-logger for all logs
- Include stack traces and execution context
- Record intention and context
For Developers: Local Setup
To develop or customize locally, clone the repository and build:
# Clone the repo
git clone https://github.com/annenpolka/roo-logger.git
cd roo-logger
# Install dependencies
npm install
# Build
npm run build
Example configuration to use your local build:
{
"mcpServers": {
"roo-activity-logger": {
"command": "node",
"args": ["/path/to/your/local/roo-logger/dist/index.js"], // adjust path accordingly
"env": {},
"disabled": false
}
}
}
Notes
- The specified directory will be created automatically if it does not exist.
MCP Tools
log_activity
— Record an activity
A tool to record an activity.
Basic example
{
"type": "command_execution",
"summary": "Run npm command",
"intention": "Update project dependencies",
"context": "Preparing for new feature development",
"logsDir": "/absolute/path/to/logs/activity"
}
Parameters
Name | Required | Type | Description |
---|---|---|---|
type | Yes | string | Activity type (command_execution , code_generation , file_operation , error_encountered , decision_made , conversation ) |
summary | Yes | string | Short summary of the activity |
intention | Yes | string | Purpose or intention |
context | Yes | string | Context information |
logsDir | Yes | string | Save directory (absolute path only) |
level | No | string | Log level (debug , info , warn , error ). Default: info |
details | No | object | Additional details (any JSON) |
parentId | No | string | Parent activity ID |
sequence | No | number | Sequence number |
relatedIds | No | string[] | Related activity IDs |
Detailed example
{
"type": "file_operation",
"summary": "Update README file",
"intention": "Clarify documentation and improve usability",
"context": "Improvements based on user feedback",
"level": "info",
"details": {
"file": "README.md",
"operation": "update",
"changedLines": 15
},
"logsDir": "/absolute/path/to/logs/activity",
"sequence": 3,
"relatedIds": ["11223344-5566-7788-99aa-bbccddeeff00"]
}
get_log_files
— List saved log files
Lists saved log files recursively. You can specify the maximum search depth.
Basic example
{
"logsDir": "/absolute/path/to/logs"
}
Parameters
Name | Required | Type | Description |
---|---|---|---|
logsDir | Yes | string | Directory to search (absolute path only) |
limit | No | number | Max files to retrieve (default: 10) |
offset | No | number | Number of files to skip (default: 0) |
logFilePrefix | No | string | Log file prefix (default: "roo-activity-" ) |
logFileExtension | No | string | Log file extension (default: ".json" ) |
maxDepth | No | number | Max directory depth (default: 3) |
search_logs
— Search saved logs
Searches saved logs with various filters.
Basic example
{
"logsDir": "/absolute/path/to/logs"
}
{
"logsDir": "/absolute/path/to/logs",
"type": "command_execution"
}
Parameters
Name | Required | Type | Description |
---|---|---|---|
logsDir | Yes | string | Log directory (absolute path only) |
logFilePrefix | No | string | Log file prefix (default: "roo-activity-" ) |
logFileExtension | No | string | Log file extension (default: ".json" ) |
type | No | string | Filter by activity type (command_execution , code_generation , file_operation , error_encountered , decision_made , conversation ) |
level | No | string | Filter by log level (debug , info , warn , error ) |
startDate | No | string | Start date (YYYY-MM-DD) |
endDate | No | string | End date (YYYY-MM-DD) |
searchText | No | string | Search text in summary or details |
limit | No | number | Max logs to retrieve (default: 50) |
offset | No | number | Number of logs to skip (default: 0) |
parentId | No | string | Filter by parent activity ID |
sequenceFrom | No | number | Sequence number lower bound |
sequenceTo | No | number | Sequence number upper bound |
relatedId | No | string | Filter by related activity ID |
relatedIds | No | string[] | Filter by any of related activity IDs |
License
MIT
Related Servers
Authless Remote MCP Server
An authentication-free, remote MCP server designed for deployment on Cloudflare Workers.
ImageSorcery MCP
ComputerVision-based 🪄 sorcery of image recognition and editing tools for AI assistants.
App Market Intelligence MCP
Analyze app data from the Apple App Store and Google Play Store for market intelligence and insights.
Deepseek Thinker
Provides Deepseek's reasoning capabilities to AI clients, supporting both the Deepseek API and local Ollama server modes.
Bifrost
Exposes VSCode's development tools and language features to AI tools through an MCP server.
Gurobi MCP
Solve optimization problems formulated by an LLM using the on-device Gurobi solver.
ComfyUI MCP Server
Integrates ComfyUI with MCP, allowing the use of custom workflows. Requires a running ComfyUI server.
Unity-MCP
A bridge between the Unity game engine and AI assistants using the Model Context Protocol (MCP).
WinCC Unified MCP Server
An MCP server for interfacing with SIEMENS WinCC Unified SCADA systems via their GraphQL API.
LLM API Benchmark MCP Server
Benchmark LLM APIs for throughput and Time To First Token (TTFT) under various concurrency levels.