Stall and Loop Check

    1

    Finds a run with no progress for 30 minutes, or the same error twice in a row, from a list of run events.

    $40

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Stall and Loop Check

    Example session with this skill installed

    Our build agents seem stuck. The run events are in events.json. Which ones have a problem, and what should I do?

    • Read your context and instructions
    • Compiled the stall and loop

    Summary

    The agent ran the check on events.json and read the two findings. Builder b1 is stalled and builder b2 is looping. b3 has no finding. The agent stopped nothing. Output below is from running the script on a small event file.

    Events

    b1  09:00 started   09:10 push   11:45 output
    b2  11:40 started   11:41 error 3fa91c20   11:50 output   11:55 error 3fa91c20
    b3  11:30 started   11:58 push
    

    Check output

    $ python3 scripts/stall_check.py events.json --now 2026-01-01T12:00:00Z
    STALL b1 170 min no progress
    LOOP b2 same error twice, sig 3fa91c20
    

    Exit code 3, because there is at least one finding.

    What to do

    BuilderFindingNext step
    b1Last progress was a push at 09:10. The 11:45 output line does not count as progress.Look at what b1 was doing after the push. Stop it or ask whoever owns it. Do not wait longer.
    b2The same error twice, with an output line between.Stop b2 and change something before trying again: the instructions, the task size or the files. A restart gives the same error.
    b3noneNothing to do.

    A stall line is a reason to look, not proof. The run may have made progress that is not recorded as an event. The events hold only a code for each error, so look up error 3fa91c20 in your own logs.

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

    About this skill

    Two things waste the most agent time: a run that stops making progress, and a run that hits the same error again and again.

    You give the script a JSON list of run events. It reports:

    • STALL: a running builder with no progress event for 30 minutes or more (you can change the limit). Output, polling, waiting and status lines do not count as progress, because a stuck run can print all day.
    • LOOP: two error events in a row with the same signature. Other events between them do not break the run. A different error does.

    Only running builders are checked, so log a started event for each one. Exit codes: 0 nothing found, 3 a finding, 2 a malformed file. The agent then says what it saw and what to do: look at the last action, stop the run, or change something before trying again.

    Standard library only. Written for Python 3.10 or later, tested on 3.13 only. 15 tests.

    What it does not do It only reports. It does not stop or restart a run or answer an approval. It cannot see events you did not record. It refuses an event time later than now. A stall line is a reason to look, not proof.

    Before you use this These skills are new and have not been tested widely. They come from a project that works and that I use. You are responsible for what you run and for reviewing what your agent does with them, so check results before you rely on them. The author is not liable for loss or damage, in line with the MIT licence.

    Requires Python 3. Cannot autonomously stop runs. Does not detect progress outside of specific logged events (started, push, state, check_result, approval, changed fingerprints).

    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

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed today

    Requires

    Python 3

    Permissions

    Terminal / Shell
    Read Files

    No network access. It reads the events file you name and writes no file. It reads the system clock once.

    Listedtoday

    What's inside

    Frequently Asked Questions

    A running builder with no progress event for 30 minutes or more. You can change the limit with --minutes. Progress means started, push, state, check_result, an approval request or answer, or a fingerprint that changed. Output, polling, waiting and status lines do not count.

    Two error events in a row from one running builder with the same signature. Other events between them do not break the run, and a different error between them does. Each repeated signature is reported once.

    A JSON list with a builder id, a UTC time, and a kind for each event. An error event also carries a short signature, which is a code for the error and never the error text. Log a started event for each builder, because only running builders are checked.

    No. It only reports. The agent tells you what it saw and what to do next. A stall line is a reason to look, not proof, because a run can make progress that is not recorded as an event.

    The script reads a JSON file, so it works with any tool that can write run events. The instructions are written for agents that read SKILL.md. I have tested it with Claude Code only.

    The SKILL.md instructions, the check script (stall_check.py), an examples file and 15 unit tests. Standard library only.

    New in LLM Ops