Weekend Creative Agent Challenge: Nightshift
Every day, an agent files a blameless postmortem for an outage that never happened. It invents the system, the failure, the timeline, and the quiet sentence at the end. EventBridge Scheduler wakes a Lambda, Bedrock writes the incident, S3 serves the page. Nobody opens anything.
Weekend Creative Agent Challenge: Nightshift
Vision & What It Does
Nightshift writes a blameless postmortem every day for an outage that never happened.
The system invents the service, the failure, the timeline, and the engineer's quiet conclusion at the end. A cache that held too long. A token pool that drained in the exact order it was designed to drain. A queue that stayed silent because silence was the configured behaviour. None of it is real. All of it reads like a document somebody filed at four in the morning.
The problem it addresses is small and specific. Postmortems are the most useful writing in engineering and the least practised. You only get to read one when something has already gone wrong at your own company, and by then you are too close to it to learn calmly. Nightshift produces a steady supply of failures that cost nothing, from systems nobody has to defend.
It is deliberately something you never open. The agent files the incident, regenerates the page, and goes back to sleep. If you want it, it is there. If you do not, nothing happened, which is exactly the point of the fiction.
The output is not a summary or a list. Each entry has a severity, a duration, a timeline of four to six moments, a paragraph explaining the actual mechanism, and one sentence an engineer would underline. The agent is also given the titles of its six most recent incidents and told to pick a different kind of system and a different failure mechanism, so the archive drifts rather than repeats.
How You Built It
Deciding what "always-on" should mean
An agent that runs continuously and an agent that runs once a day are the same thing from the reader's side and very different things on the bill. I picked scheduled execution because the challenge prompt already answered the question: the best tool is the one you never have to open. Something you never open does not need to be awake.
That decision removed a server, a container, and a websocket from the design before I wrote any code.
Writing the prompt before writing the infrastructure
I ran the reasoning prompt through the AWS CLI first, against Bedrock directly, before creating a single resource:
aws bedrock-runtime converse \ --region us-east-1 \ --model-id us.amazon.nova-2-lite-v1:0 \ --messages '[{"role":"user","content":[{"text":"..."}]}]'If the model could not invent a failure that sounded like it came from a real incident channel, there was no product. It could. The instruction that made the difference was negative rather than positive: no heroics, no villain, the system did exactly what it was told. Without that line the output drifted into thriller. With it, the tone settled into the flat regret that real postmortems have.
Nova wraps its JSON in markdown fences no matter how firmly the system prompt forbids it. Forcing the shape through Converse toolUse would be the correct answer. Stripping the fences on the way out was the fast one, and this build was optimised for fast.
Giving the agent a memory without giving it a database
The agent reads its own archive from S3 before generating, and the six most recent titles go into the prompt as things to avoid. That is the entire memory system. No DynamoDB table, no vector store, no embedding call. A JSON file it overwrites once a day.
The same Lambda run also regenerates the whole HTML page from that file, so the site has no build step and no framework. Rendering an archive of sixty entries as a string is cheaper than deploying anything that could render it properly.
Proving it runs on its own
The requirement is not that the agent can produce output. It is that it produces output without anyone starting it. So after deploying the daily schedule I temporarily set it to
rate(3 minutes) and walked away.entries: 3 2026-08-21 20:26 UTC | SEV2 | Stale Cache Propagation 2026-08-21 20:23 UTC | SEV2 | Ephemeral Token Exhaustion 2026-08-21 20:22 UTC | SEV2 | Silent Queue BuildupThe first line is the manual invocation that seeded the archive. The two above it are EventBridge Scheduler, unattended. Each page carries the generation timestamp, so the evidence is on the live site rather than in a screenshot.
AWS Services Used / Architecture Overview
Service | Role |
|---|---|
Amazon EventBridge Scheduler | Wakes the agent once a day |
AWS Lambda | Node.js 22, the whole agent, no framework |
Amazon Bedrock | Amazon Nova 2 Lite, via the Converse API |
Amazon S3 | Archive, memory, and static website hosting |
Amazon CloudWatch Logs | Execution record for each unattended run |
EventBridge Scheduler (cron, 21:00 UTC daily) | vAWS Lambda --read--> S3: archive.json (the agent's memory) | +--invoke--> Amazon Bedrock, Nova 2 Lite | +--write--> S3: archive.json + index.html | v S3 static website (public, no CloudFront)There is no API, no auth, and no client-side JavaScript. The page is a string the Lambda writes. CloudFront was left out on purpose: it adds a cache to invalidate and a distribution to wait on, and this site changes once a day.
The whole thing fits inside the Free Tier apart from Bedrock, and one short generation a day puts the model cost below the rounding error on a monthly bill.
What You Learned
The scheduled agent is a different design problem from the interactive one, and the difference is not the trigger. It is that nobody is waiting. There is no loading state to design, no error to surface to a user, no latency budget. What matters instead is that the output is worth finding later, which pushes all the effort into the prompt and the archive rather than the interface.
Giving the agent its own recent output as an input turned out to be the entire personality of the thing. The first version, with no memory, produced three database outages in a row. Passing back six titles with an instruction to diverge changed the archive from a loop into a drift. That is a much cheaper mechanism than it sounds like, and I had assumed it would need a store of some kind before it would work at all.
Proving autonomy is a build requirement, not a documentation step. I nearly shipped this with a daily schedule and a screenshot of the code. Temporarily setting the schedule to three minutes cost nothing and turned a claim into a live artefact, and the timestamps sitting on the public page are stronger evidence than anything I could have written about them.
Writing fake postmortems for a few hours also made something obvious that I had not noticed in the real ones. Almost every failure the model invented was a system doing precisely what it had been configured to do, at a moment when that was the wrong thing. It was told not to write villains, and what filled the gap was configuration.
Link to App or Repo
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article