More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    agent time tracker

    1

    Honest time tracking for AI agents. Start a task timer before work, stop it when done, and get per-project and per-task reports even with several agents running in parallel.

    Free

    Security scanned
    agent time tracker

    agent time tracker

    Example session with this skill installed

    Start a new task called "Refactor Auth Logic" for the current repository and let me know when it's running.

    • Read your context and instructions
    • Compiled the agent time tracker

    I have initialized the timer for your new task.

    Task Started

    PropertyValue
    Task TitleRefactor Auth Logic
    Project/Repohttps://github.com/user/auth-service.git
    StatusRunning
    Start Time2024-05-20 14:30:05

    Next steps

    • Proceed with the authentication logic refactor.
    • Tell me "Task done" once the refactor is complete to stop the timer.
    • Ask for a "status report" if you need to see all active timers across this project.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Accurately track how much time AI agents spend on specific tasks and projects. This skill addresses the lack of visibility into agentic labor, allowing developers to generate honest timesheets for client billing or internal productivity analysis. It handles parallel agent sessions without double-counting, distinguishing between wall-clock real time and total combined session time.

    What it does

    • Task management starts and stops timers for specific units of work like bug fixes or research.
    • Project grouping organizes time entries by repository URL, working directory, or project name.
    • Reporting generates summaries for specific periods, including real-time and total combined session hours.
    • Manual logging adds entries for non-agent work like meetings or calls directly into the timeline.
    • Parallelism tracking monitors multiple simultaneous Claude sessions and calculates the parallelism factor.

    How it works

    1. Initialize by starting a timer via timer_task_add or timer_task_start before any work begins.
    2. Contextualize every call by passing repo or project identifiers to ensure time is logged to the correct client.
    3. Finalize by calling timer_task_done immediately upon completion of the unit of work.
    4. Query the agent for status or reports using commands like timer_report or timer_status.

    Frameworks & tools

    Uses the Timer MCP connector (timer.kopachev.life/mcp) for data storage and retrieval. Integrates with Claude Code hooks and Git for automatic repository identification.

    Why this beats prompting it yourself

    Prompting an agent to "remember the time" fails because LLMs lack an internal clock and context of parallel runs. This skill enforces a strict start/stop protocol backed by a persistent MCP server, ensuring billable minutes are never lost or estimated.

    Use cases

    • Generating billable hour reports for freelance clients based on repository activity.
    • Auditing agent efficiency by comparing wall-clock time against total session minutes.
    • Managing long-running background tasks across multiple browser or terminal sessions.
    • Correcting manual time entry errors or moving logs between different project folders.

    Known limitations

    Entries shorter than sixty seconds are not stored. MCP tool responses and the web interface currently use Ukrainian and require translation by the agent. Continuous thought blocks exceeding 20 minutes may trigger early entry closure.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Productivity.

    Entries <1 min not stored. MCP tool responses are in Ukrainian. 20+ min of model silence may auto-close entries. Requires Timer MCP server.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 21 days ago

    • Free with an Agensi account

    Accounts
    GmailGitHub
    Software
    Git

    Permissions

    Terminal / Shell
    Read Files
    Network Access
    Browser

    Allowed Hosts

    gmail.googleapis.com
    api.github.com
    Listed21 days ago

    What's inside

    Frequently Asked Questions

    This skill solves the 'internal clock' problem in LLMs by providing a persistent MCP server that tracks exact start and stop times, preventing the agent from guessing or forgetting billable duration during complex tasks.

    Yes, the skill is designed to handle parallel sessions; it distinguishes between 'wall-clock' time (the actual duration passed) and 'session time' (the combined effort of all active agents), ensuring you don't double-bill clients for simultaneous tasks.

    The skill requires the Timer MCP connector (timer.kopachev.life/mcp) and is optimized for Claude Code and Git-integrated environments for automatic repository and project identification.

    The skill includes the logic for task management, project grouping, and reporting, though users should note that current tool responses are in Ukrainian and are automatically translated by the agent for the user.

    Simply initialize the Timer MCP connector in your agent's environment and use the 'timer_task_start' command to begin logging; the skill will automatically use repo URLs or directories to categorize the work.

    The system ignores entries shorter than 60 seconds to prevent log clutter, and continuous thought blocks exceeding 20 minutes may trigger an automatic closure to ensure data integrity during long-running background tasks.

    New in Productivity

    Free