EXTENSIONS

Render Markdown files into a static HTML site with an index page, without requiring Python or special content markers.

Direct-upload Cloudflare Pages deployment and custom-domain binding for sites that do not use a Git repository.

Run pixiecore as a managed container, auto-configured to serve netboot.xyz to any host that netboots on the LAN.

Inventory, assess and safely replace SSH private keys across every host your control account can reach, with an HTML report and per-key remediation advice.

Inventory and back up DreamHost shared-hosting accounts over SSH, and manage DreamHost DNS records through the official API.

Generic systemd *user* service management: idempotently create, start, stop, restart, and remove a persistent user service for any command line — and audit every service model against the live host in one call.

Find every ebook on your machine, detect its bibliographic metadata, resolve authors and titles against Wikipedia + Wikidata, and render a browsable HTML page linking each file to its location on disk.

A Wikipedia domain parser that reads cached responses from the shared cache

A Wikidata domain parser that reads cached responses from the shared cache

A managed, disk-backed HTTP GET cache — the single place responses are fetched, cached, paced, and retried, so other models read the cache instead of re-implementing (and re-hitting) the network.

The release dashboard for a swamp extension repository. It draws the dependency graph of every extension you own plus the extensions they use, shows what needs publishing (and in what order, and to which channel), and reports release-hygiene compliance and test depth — so "what do I ship next?" is one command.

The documentation quality gate for swamp extensions. Enforces one guiding principle — the published manifest plus README must be enough for a user or agent to easily learn how to do everything the published extension does.

A tracked curl wrapper: every HTTP request records exactly what went out and what came back — headers, cookies, body, redirects and timing — with secrets masked, and a persistent cookie jar that carries a session across requests.

Run GGUF models that need a customised llama.cpp kernel — the ones stock llama.cpp (and therefore Ollama) cannot load — by installing a patched build, serving them with `llama-server`, and verifying they answer.

The full news stack: feed discovery, catalog management, preference-aware fetching, LLM story fusion, and static HTML rendering — plus a decoupled feedback server that closes the 👍/👎 loop between the rendered page and the workflows.

Caddy reverse-proxy and service management for a Linux host with systemd.

============================================================================ ⚠️ SECURITY WARNING — THIS CAN EXPOSE MAGNITUDE ON THE NETWORK ============================================================================

Turn "what models will help me do <something>?" into a ranked, hardware-aware shortlist from the Hugging Face Hub — then download the one you choose, at a pinned revision, ready for an Ollama Modelfile to `FROM`.

A swamp vault backend that stores secrets encrypted at rest with `systemd-creds --user`, so the encrypted files are useless without your UID and this machine's id. No root, no desktop keychain daemon, no unlock step.

Sample every host in a swamp fleet — disk, network IO, load and memory — and push each sample to the swamp-managed OpenTelemetry system, every 15 minutes.

Operate a Sipeed NanoKVM from swamp: a health verdict, screenshots, GPIO power/reset and virtual media, with the device's health pushed to your swamp-managed OpenTelemetry collector.

Enlist a host as a swamp worker for one `swamp serve` orchestrator: install or upgrade it, reconcile its systemd user unit, mint its enrollment token, and report or remove exactly the worker this repo owns.

Everything Ollama lets you do with a model, as swamp methods — and a Modelfile generator that customises a model to this host's GPU and RAM, then builds it and proves it loads.

This host's accelerators as swamp data — every GPU (vendor, name, VRAM, driver, compute capability), the total VRAM, system RAM, usable CPU cores and the compute runtime (CUDA / ROCm / Metal / none) — read straight off the machine with no network and no SSH.