Skip to content

ericrisco/rsc-harness / skills/hetzner

hetzner

Unverified

Use when provisioning or hardening a Hetzner Cloud VPS: picking the plan/location (CX/CPX/CAX/CCX), bringing it up reproducibly with hcloud + cloud-init, locking down SSH, and wiring the Cloud Firewall to hand off a Docker-ready host. NOT app deploys on Coolify (that is `coolify`), NOT Dockerfiles (that is `docker`), NOT another provider (that is `digitalocean`).

FileKindSize
SKILL.mdskill md10.3 KB
evals/README.mdmarkdown1.8 KB
evals/cases.yamlconfig3.6 KB
references/cloud-init.mdmarkdown4.3 KB
references/plans-and-locations.mdmarkdown2.9 KB
scripts/verify.shexecutablescript3.9 KB

evals/README.md

View on GitHub
# Eval harness — hetzner

`cases.yaml` is the fixture; this file is the procedure. The evals are run by an
**agent harness** (an agent with the full skill catalog loadable on demand), not
a pure script. Two things are measured: **triggering** (does the skill fire on
Hetzner provision/harden/plan/firewall prompts and stay quiet for sibling
providers and app-deploy) and **capability** (does loading it produce a genuinely
reproducible, hardened flow rather than a console walkthrough).

## Triggering

For each `should_trigger` / `should_not_trigger` item: start a fresh session with
the full catalog, feed the `prompt` verbatim, record which skill the agent
invokes, and run 3–5 trials (the choice is stochastic). Pass when **hetzner** is
chosen for the majority of `should_trigger` trials and is *not* chosen for
`should_not_trigger` (ideally the agent routes to the listed `route_to` sibling).
Target >= 90% trigger accuracy across prompts.

## Capability

For the `capability` scenario run two arms — **WITH** only hetzner loaded vs
**WITHOUT** any skill — three times each. Grade each response against
`must_include` (one point per item that is genuinely present and correct, not
hand-waved). Pass when WITH covers >= 80% of the rubric and beats WITHOUT by a
clear margin (target >= 25 points). A skill that doesn't move the needle fails
even if the baseline answer was decent.

## Honesty notes

These are stochastic, LLM-graded evals — re-run on edits and treat small deltas
as noise. `route_to` targets (coolify, digitalocean, docker, domains-dns, fly-io)
assume those siblings exist in the catalog; a missing sibling can cause a
near-miss mis-route that isn't a hetzner fault — note it, don't count it against
the skill. A static lint of the emitted cloud-init artifact is available
separately via `scripts/verify.sh`.
writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deploymentStars: 322kUpdated Claude CodeCodexVerifiedInstall
xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release.Stars: 118kUpdated Claude CodeCodexVerifiedInstall