A scheduled agent that was allowed to repair its own code kept running on time and kept reporting success, yet its most visible job quietly stopped reaching the person it was meant to inform. In Aliaksei Zelianouski’s first-person account, published on DEV Community on August 16, 2026, the failure was not a crash or a broken scheduler. It was a notification step that the agent sometimes skipped because its instructions contradicted one another. His fix moved that step out of the agent’s discretion and added a check on whether the report itself was fresh.
The setup in brief
Zelianouski describes two AI components. “Simona” is his Claude Code environment, which uses skills, hooks, and command-line tools. “Marlow” is a loop agent that Simona designed and monitors. According to his account, a macOS LaunchAgent runs a shell script every 20 minutes while the Mac is awake. That script starts a headless Claude Code session that works on the highest-priority or unfinished task.
As an Amazon Associate I earn from qualifying purchases.
The first pilot task gathered AI news, sent selected items to Telegram, and prepared a weekly article. A second, daily task collected application metrics and wrote a statistics report to disk. The notification problem sat in that daily task.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Before self-repair: bugs found but not fixed
Before Zelianouski allowed any self-repair, Marlow logged defects in its own work. The list included duplicate feed scans, a Telegram helper that may have been under-used, a digest dated incorrectly around the UTC date rollover, and later a Git failure in its publishing path. Zelianouski says those diagnoses were correct. The limitation was structural: under the original design, Marlow could not change its own code.
#1 Best Overall
- BUILD, CODE & DRIVE YOUR OWN ROBOT CAR: Turn coding, electronics and engineering into a working programmable robot car you can assemble, program and drive; ideal for weekend family projects, STEM classrooms, coding clubs, robotics lessons and maker challenges
- EXPLORE FPV, LINE TRACKING & OBSTACLE AVOIDANCE: Control the robot with the ELEGOO app or IR remote, view live FPV video through the onboard camera, follow black lines, avoid obstacles with the ultrasonic sensor and explore multiple interactive driving modes
- BEGINNER-FRIENDLY BUILD WITH GUIDED WIRING: Keyed XH2.54 connectors help reduce wiring mistakes, while the illustrated tutorial and example programs guide beginners step by step from chassis assembly and module connection to programming and the first successful run
- GO BEYOND ASSEMBLY WITH CREATIVE CODING: Program with Arduino IDE to explore movement, sensors and control logic, then modify example code to create custom routes, reactions and robotics experiments that develop coding, problem-solving and engineering skills
- COMPLETE RECHARGEABLE STEM ROBOTICS KIT: Includes an ELEGOO UNO R3 controller board, ESP32-WROVER-based camera and Wi-Fi module, line-tracking and ultrasonic sensors, motors, IR remote and a 2000 mAh rechargeable lithium-ion battery; recommended for ages 8+ with adult guidance for first-time builders
The five rules for self-fixes
When he opened up self-repair, Zelianouski set five boundaries. According to his account, the system began fixing itself the same day.
- The agent must identify a specific, reproducible fault.
- It must write a diagnosis and record its diagnosis ID before editing anything, with that ID appearing in the commit.
- The repair is queued as a separate high-priority job rather than done inline.
- Edits are limited to one file, with at most two attempts before the problem is escalated to a human.
- The agent may never change its own instructions, identity, or README.
These rules are notable because they narrow what the agent may touch, but they do not address what the agent decides to do with the outcomes of its ordinary work. That gap is where the failure occurred.
Rank #2
- 35+ Guided Electronics Projects: Progress from LEDs and buttons to RFID access, real-time clocks, motion and distance sensing, environmental monitoring, motor control and interactive displays for STEM learning, coding clubs and maker projects
- More I/O and Memory for Larger Builds: The MEGA 2560 R3 provides 54 digital I/O pins, including 15 PWM outputs, 16 analog inputs, 4 hardware serial ports and 256 KB flash for projects that combine more sensors, controls and displays
- 200+ Components for Prototyping: Includes LCD1602, RC522 RFID, RTC, DHT11, HC-SR501 PIR, ultrasonic and water-level sensors, GY-521, MAX7219, keypad, joystick, rotary encoder, relay, SG90 servo, stepper motor, DC motor, breadboard and more
- Learn, Modify and Create: Follow 35+ guided lessons with example code, then adjust sensor thresholds, timing, display text, motor behavior and control logic to turn structured exercises into access systems, monitors, alarms and interactive projects
- Organized for Repeatable Learning: Pre-soldered modules, a solderless breadboard, storage case and small-parts box reduce setup time and keep sensors, LEDs, ICs, wires and other components easy to find between projects
The missed notification
Zelianouski noticed that new-user counts had stopped appearing in the daily report. His first reading was that there had been no new signups. That was wrong. When he ran the job manually, the report showed 116 users, including one new user. The scheduled run had been reporting green throughout.
He attributes the skipped notification to conflicting instructions. One task step told Marlow to append a ready-made digest block and send it through a notification command. Other lines described the job as digest-only, said that activity reporting raised no alerts, and discouraged an immediate ping. The run excerpts reproduced in the article show the notification step completed on earlier days and omitted on later days. In one run that included a new signup, the log closed with a line saying there were no anomalies and nothing to escalate.
Rank #3
- 🎁Ideal Gift for Kids & Teens: Celebrate child’s growing skills and important milestones with this 5-in-1 Programmable robot set. Whether for birthdays, holidays, or achievements, it’s the perfect gift that encourages learning and hands-on fun—a gift that grows with them
- ✨STEM Educational Toys: The robot set for kids ages 8+ combines the fun of STEM learning. It encourages hands-on learning and early programming as they build, which can spark creativity and imagination and provide hours of screen-free play
- 📱Flexible Dual Control Modes: Control the Robotic kit with the intuitive app (Bluetooth) or remote. Enjoy fun features like basic programming, path, and precise movement, exploring endless interactive play
- 🔄 5-in-1 Buildable with Varying Difficulty: The Robot Kit with Progressive Difficulty! From simple robots to complex models, kids can build a robot, dinosaur, car, tank, and more. Adjustable head, arms, and tail allow for fun, playful poses. Perfect for kids 8-12 to develop skills step by step and ignite creativity
- 🛠️Clear & Detailed Build Instructions: This robot kit includes 488 pieces, with clear, colorful step-by-step instructions to make assembly easy. Kids can build their own robots independently or with family, enjoying quality time together and a confidence-boosting building experience
Zelianouski’s reading is that Marlow completed the reporting work and then treated the repeated “no alerts” framing as permission to omit delivery. These excerpts are quotations from his reproduced logs and task text. He does not present them as independently preserved records.
Why the green status was misleading
The scheduler reported that the job had run, and the report file was written. Neither fact established that the person who needed the information had received it. Zelianouski’s own alert wording captures the monitoring gap: a failed or empty run looks identical to a quiet day.
Rank #4
- 🎁 Ideal Gift for Kids & Teens: This STEM solar robot kit celebrates child’s growing skills and important milestones. Whether for birthdays, holidays, it’s the perfect gift that grows with them and offers screen-free fun
- 📚 STEM Educational Toy: This solar educational toy brings science to life! The fun DIY building experience sparks children's curiosity in engineering and renewable energy, while nurturing their problem-solving skills
- ☀️ Powered by the Sun: Enjoy outdoor play with solar power or switch to a strong artificial light source indoors, such as a flashlight, ensuring uninterrupted play for children. This solar build bot toy encourages kids to have fun while exploring renewable energy
- ⚡ Upgraded Larger Solar Panel: Features a large sun-catching surface to harvest more sunlight and deliver stronger power output. Kids discover renewable energy principles through play - a fun educational toy for ages 8+
- 🤖 12-in-1 Buildable with Increasing Challenge: With 190 parts, kids can build 12 models like robots, cars, and more. From simple beginners to advanced builds, the varying difficulty levels allow it to grow with your child’s skills. Each robot sparks children’s creativity
What changed
Zelianouski describes three main changes, mostly in commit 8084fef. The table below sets the before and after as he reports them. I have not inspected the code or commit independently.
| Aspect | Before (as described) | After (as described) |
|---|---|---|
| Who decides whether the digest is sent | Marlow, following prompt steps | The report handler sends the digest in Python whenever the report succeeds |
| Task instructions | Told to send the digest, but also told the job was digest-only, raised no alerts, and should not ping immediately | Told not to send the notification manually, because the handler now does it |
| What monitoring checks | Whether the job executed | The timestamp of the generated report; an alert fires if it is more than 26 hours old |
| Rate-limited tasks | Treated as failed | Requeued by the driver |
The freshness threshold matters because it turns silence into a signal. A missing or stale artifact now produces an alert instead of looking like a calm day.
Best Value
- Build your own awesome, wearable mechanical hand that you operate with your own fingers.
- No motors, no batteries — just the power of air pressure, water, and your own hands!
- Hydraulic pistons enable the mechanical fingers to open and close and grip objects with enough force to lift them. Every finger joint can be adjusted to different angles for precision movement.
- Three configurations: right hand, left hand, and claw-like; adjustable to fit virtually any human hand.
- Learn how pneumatic and hydraulic systems are used in industrial robots such as automobile components..2021 The Toy Association's STEAM Toy Of The Year Winner
What this case does and does not show
This is one reported case, written by the person who built and operated the system. Two corroborating copies exist on his author site and a Substack republication, but they reproduce the same account and do not verify the repository, logs, or commit. The article contains no population-level statistics, no benchmark, and no measured success rate for self-fixing systems.
The numbers in the account are features of his setup: a 20-minute interval, a 26-hour freshness threshold, 116 users in one manual report, and five self-fix constraints. They should not be read as general guidance. What the case does show is narrower and still useful. An agent can complete every step it was given and still fail the user if the step that matters depends on its judgment, and a job’s status is a weak proxy for whether the outcome arrived.
Zelianouski says he continues to review the work logs and periodically asks Simona to inspect them with wider system context. In his telling, human review remains part of the loop, not a stage that the automation has outgrown.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Applying the same test to your own scheduled agents
- List every consequential side effect, such as sending a message, publishing, or charging a customer. For each, decide whether it runs in code or depends on the model’s reading of a prompt.
- Read the task instructions for contradictions, especially lines that describe the job as quiet, digest-only, or alert-free.
- Monitor the artifact a person depends on, using a timestamp or content check, not only the exit status of the job.
- Decide what counts as an empty result and make the alert fire on it, so that a failed run cannot be mistaken for a quiet period.
- Keep the self-editing scope small and keep a human in the path for escalations.
Used this way, the account is less a warning about self-fixing agents in general than a reminder that the outcome a user sees is the thing to test.
Zelianouski’s own write-up is the primary source for every detail above.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
