30 posts shown
CVE volume is accelerating and most teams are running scripts that can't keep up. We used an agent to build a scanner that queries three databases, classifies and scores every CVE on actionability.…
Useful links and daily AI-assisted news digests.
Useful links and daily AI-assisted news digests.
I've had a few people ask me, in roughly this order, whether AI is going to replace incident response, then whether I've actually tried it, then whether it's any good. So here's the honest answer,…
Why the autonomous agent loop inside the Firecracker cage is a deterministic state machine and not a swarm: how gobrr decomposes, schedules, and gates the work, plus what held and what bit me…
Stop Marrying Your Infrastructure Tools
Infrastructure was the first use case we looked at with swamp. Then users started solving security, finops, and validation problems with the same model. By the fifth time, it was hard to treat it as…
Why I built a Firecracker microVM extension before the autonomous agent loop that needs it: the lethal trifecta, the painful microVM build, and a loop that shipped a real swamp extension end to end.
Integration practice for the manual work that should already be software. AI systems, internal tools, and the integrations that take the human out of the middle of workflows that should run on their…
Why the ABN affordability number isn't the real number for an expat in the Netherlands — and how I turned a fuzzy apartment-and-mortgage decision into a plan I can actually steer, with good-planning…
How I turned a noisy Amsterdam apartment feed into a clean, high-signal one I can reason with — good-planning underneath, a move.nl dossier on top, public-data enrichment, and my requirements encoded…
Day three with Swamp: The first commands and interactions.
Most SRE work follows a pattern. An alert fires, you check the metrics, pull the logs, form a theory, fix it. Then you do the same thing next week. That sequence is a workflow. When agents can…
A developer describes using an AI assistant (Hermes) running on a homelab to automatically monitor Todoist tasks and transcribe conference recordings into Obsidian notes, reducing the "capture debt"…
Declarations were scaffolding for blind agents. Ceremonies were scaffolding for blind organizations. The same move removes both.
A developer describes replacing a manual pre-deploy checklist with automated HTTP-based E2E tests using the Swamp framework and a custom TypeScript/Deno extension, eliminating the need for Cypress or…
Every IaC tool promised "any system." Few delivered once you tried to use it in your own business. Provider ecosystems were designed for humans who can fill in the gaps. Agent ecosystems need a…
Benchmark results from an M5 Max Mac show that local LLMs running via Ollama and OpenCode — particularly the 35B Qwen 3.6 MoE model — are fast and reliable enough for bounded agentic tasks, though…
The governance you know is the governance you repeat.
Every company has this person. The engineer who every production change flows through because they're the only one who knows the constraints. If you're thinking about putting AI agents anywhere near…
Deming told us. We made the signature the proof and called it quality.
On May 30th I published v0.1 of a Databricks integration pack for Swamp. By the end of the same evening it was at v0.13. Fifteen models, 100% A on every release, every model end-to-end smoke-tested.…
A developer documents the environment-snapshot mechanism in the Hermes AI agent runtime that caused tool subprocesses to inherit stale API keys, and identifies /opt/data/home/.bash_profile as the…
What a development system looks like when you build it to keep yourself honest
A homelab engineer describes using an AI agent (Amp) alongside a declared Swamp model architecture to turn a Frigate disk-full alert into an automated, rate-aware notification that includes projected…
A developer used Claude Code to build an offline meeting transcription pipeline on the Swamp framework, combining mlx-whisper, pyannote diarization, and a local Ollama LLM to produce Granola-style…
The sovereign cloud movement is forcing a question the industry has been avoiding. When you have to move off a hyperscaler, the bottleneck isn't the new provider's APIs. It's how precisely you can…
A developer describes building ADW, a 10,000-line Ruby pipeline that automates code shipping against a production Rails monolith through phased AI agents, and explains a migration to improve…
AI agents are probabilistic. The same prompt doesn't always produce the same output. That's not a problem to avoid. It's a problem to engineer around. Typed schemas, validated execution,…
Drawing parallels with Linux packaging ecosystems, this post proposes a 'swamp forge' to make collaborative extension development the default rather than the heroic choice.