Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    wdog

    1

    A watchdog runner that monitors long-running shell commands for hangs, prompts, and orphan processes.

    Free

    Security scanned
    wdog

    wdog

    Example session with this skill installed

    Run a clean install and build for this npm project. It usually takes 5 minutes. If it hangs or asks for input, stop it.

    • Read your context and instructions
    • Compiled the wdog

    I will execute the install and build process using wdog to monitor for hangs or interactive prompts.

    Build Execution

    
    # Starting the build as a named job to monitor progress
    wdog start --name project-build --max 15m -- 'npm ci' 'npm run build'
    
    # Waiting for completion with an 8-minute poll interval 
    # (exit 124 will trigger a retry if the job is still healthy)
    wdog wait project-build --for 8m
    

    The command initiated a supervised process tree. wdog is monitoring for STALL alerts (120s of inactivity) and PROMPT detection.

    FieldValue
    Job Nameproject-build
    Max Timeout15m
    StatusRUNNING
    Stepsnpm ci, npm run build

    Next steps

    1. Run wdog status project-build to check the current step and runtime.
    2. Use wdog tail project-build -n 50 if you see an ERROR_LIVE alert.
    3. The processes will be automatically cleaned up if a stall is detected.

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

    About this skill

    Long-running shell commands often fail silently, hang on hidden prompts, or leave orphan processes that eat resources. Plain shell redirection and manual polling make it impossible to track the health of a build or a remote migration without constant babysitting.

    What it does

    • Watchdog execution monitors process trees for CPU stalls, I/O activity, and hung states.
    • Prompt detection identifies when a command is waiting for user input and fails fast to prevent hangs.
    • Job management handles background processes with named tracking, status monitoring, and cleanup.
    • Remote orchestration executes and monitors commands on remote hosts over SSH with automatic self-installation.
    • Port/URL polling waits for dev servers or services to become healthy before proceeding.
    • Log analysis extracts error patterns and summarizes failures without dumping thousands of lines of noise.

    How it works

    1. Wrap commands using wdog run -- 'command' for immediate tasks or wdog start for background jobs.
    2. Monitor alerts by checking for STALL, PROMPT, or ORPHAN lines in the output stream.
    3. Control execution with wdog wait to block until a specific port is open or a timeout is reached.
    4. Inspect failures using wdog tail or wdog errors to see exactly why a process tree stopped.

    Frameworks & tools

    This tool is a Linux-compatible binary that works with any shell command, including Docker, Node.js, Rust, Python, and SSH. It integrates with system-level process tracking and requires no root privileges.

    Why this beats prompting it yourself

    Writing robust shell wrappers for timeouts and orphan cleanup is error-prone and usually misses edge cases like zombie processes or interactive prompts. This skill provides a standardized protocol for process supervision that handles the lifecycle and failure modes of complex dev ops tasks out of the box.

    Use cases

    • Supervise long CI/CD build pipelines and test suites to prevent runner timeouts.
    • Orchestrate dev server startup sequences that depend on database migrations.
    • Manage remote database maintenance tasks over flaky SSH connections.
    • Clean up leaked processes from complex integration tests automatically.

    Known limitations

    Supports Linux only (x86_64). Requires a binary installation on the host or remote machine. Standard input is closed by default to prevent silent hangs on prompts.

    Linux x86_64 only. Requires binary installation. Closes stdin to prevent hangs, so interactive commands must be run in non-interactive modes.

    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 any payment.

    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 forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 10 days ago

    • Free with an Agensi account

    Requires

    GitHub
    Linux
    Git

    Permissions

    Terminal / Shell
    Network Access
    Read Files

    Allowed Hosts

    raw.githubusercontent.com
    api.github.com
    github.com
    Listed10 days ago

    What's inside

    Frequently Asked Questions

    Unlike standard shell timeouts, wdog detects "silent hangs" like interactive prompts or CPU stalls, and it automatically cleans up orphaned child processes that would otherwise continue to consume system resources.

    It is designed for any AI agent capable of executing Linux shell commands and can be integrated into custom CI/CD pipelines, DevOps workflows, or local terminal environments.

    The skill is a Linux-compatible (x86_64) binary that requires no root privileges; it includes a self-installation feature for remote hosts via SSH to simplify multi-server management.

    By default, wdog closes standard input to force commands that require user interaction to fail fast, allowing your automation to handle the error immediately rather than hanging indefinitely.

    You receive the core wdog binary, documentation for all subcommands (run, start, wait, tail), and the logic required to monitor complex process trees across local and remote environments.

    New in Workflow Automation