30 posts shown
You can’t deny Swamp’s gang speed in addressing users requests.
Swamp is an automation framework built to be operated by AI agents rather than by people writing pipelines. What it is, and how I run my autopilot on it.
I taught swamp to read Fedora (accounts, Bugzilla, the packager dashboard) so one command tells me what my packages need. The answer was 135 things.
A developer describes building a route-risk monitoring system using Hermes and Swamp during a multi-state road trip, where structured models and a change-detection loop provided weather and hazard…
Engineers have been using code they didn't write for decades — NPM packages, Stack Overflow pastes, codegen they never read. Nobody called it slop. Then the author changed...
A developer recounts using the Fable 5 model to diagnose and resolve a compounding homelab automation failure — a monolithic 95MB S3 index, orphaned write locks, and wedged processes — through…
A developer describes building a CLI agent tool that persists structured records of each invocation — tracking model, cost, token counts, retries, and wall-clock time — to enable meaningful…
A developer describes building a Pixel Watch complication pipeline that pulls two specific Home Assistant entities — homelab status and Swamp Club score — through MQTT and the Companion App, with…
Writing code was always the surface. The real work was the frameworks and processes underneath. Machines have the typing now; what's left is what always mattered: context, and the deterministic…
The dominant pattern in agentic automation is ephemeral agents producing deterministic artifacts. What does that architecture leave on the table?
I modeled my newborn as a typed resource with a sync method and a stack of reports. It's the newest of five ways I track one baby, and somewhere in the absurdity there's a real reason two exhausted…
I taught swamp to drive FreeIPA, then used it to issue my laptop an 802.1X client cert and get onto my own network. Mostly this is about knowing when to write things down.
A result that travels the way a record travels: a stamped final result, a fingerprint anyone can check, and the full chain that made it attached.
Six months into 2026, the difference between teams using AI effectively and teams still evaluating it isn't a head start anymore. It's a different way of working entirely. And the gap isn't just…
How we built @twonines/fact-store — a swamp extension that gives AI agents persistent, adversarially-reviewed operational memory.
Useful links and daily AI-assisted news digests.
If you’re building agentic systems, it’s tempting to use stochastic AI for all the steps in your process.
The retirement of Chef Infra Server's open-source edition prompts a reflection on configuration management's evolution, arguing that Kubernetes absorbed—rather than eliminated—the convergence loop…
A software engineer reflects on how LLM-based agentic development represents a paradigm shift comparable to infrastructure-as-code, and introduces Swamp, an open-source tool built by Chef co-founder…
Recognizes who is present, then written rules set the room — explainable and recorded every time.
Paul Stack shares the SWAMP team's real-world experience of using multiple agents to refactor a software product, pushing incremental changes to users the whole way.
I could have Claude Code directly create infrastructure for me. But I prefer to have it write infrastructure code, for consistency, confidence, and composability.
Useful links and daily AI-assisted news digests.
Zero API calls. Still found the security group rule you added.
How I use AI with Swamp reports to review pull requests, learn unfamiliar codebases, and reach better engineering questions sooner.
How I evolved a disposable VM lab into a Swamp-managed infrastructure controller using VM pools, capability planning, and data-driven workflows.
A workflow named generate-ixens on abnormalia.com describes itself as a DAG-shaped process that generates "ixens" from a YAML input file, explaining its own job dependencies, parallel execution, and…
Useful links and daily AI-assisted news digests.
The LLM does not create the thinking gap. It makes the gap visible and consequential for the first time.
The post argues that sandboxing AI ops agents is counterproductive and proposes a "playground" paradigm instead, granting agents autonomy within bounded infrastructure, pre-vetted workflows, and…