nixos-config: Skill for Claude Code

.agents/skills/nix-verification-loop/SKILL.md

nix-verification-loop is a skill for Claude Code, Codex from FilipNowakowicz/nixos-config. It costs 29 tokens per session (2,487 once invoked), scanned B, original, MIT.

A validation workflow for a NixOS project, where NixOS is a Linux system configured with reproducible code. It selects a focused check based on whether the change affects the system setup, a package, a host, a profile, or documentation.

In plain words
What is it for?
Use it when changing Nix files, shell scripts, hosts, packages, tests, deployment wiring, secrets boundaries, generated data, or project documentation.
Why use it?
It helps developers run an appropriate first check instead of manually choosing among many validation commands.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one. Also seen: mentions CLAUDE.md; mentions subagents; installed under .agents/ (shared by several agents).

This is FilipNowakowicz/nixos-config's own configuration. It tells Claude Code and Codex how to work on nixos-config itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything nixos-config configures →

Reuse

Borrowing it

Nothing to install: this file belongs to FilipNowakowicz/nixos-config. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/FilipNowakowicz/nixos-config/main/.agents/skills/nix-verification-loop/SKILL.md
Clone the repo
git clone --depth 1 https://github.com/FilipNowakowicz/nixos-config

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for nix-verification-loop

README.md
[![agentmods](https://agentmods.dev/badge/skills/filipnowakowicz/nixos-config/nix-verification-loop/github.svg)](https://agentmods.dev/skills/filipnowakowicz/nixos-config/nix-verification-loop)
Your own site
<a href="https://agentmods.dev/skills/filipnowakowicz/nixos-config/nix-verification-loop"><img src="https://agentmods.dev/badge/skills/filipnowakowicz/nixos-config/nix-verification-loop/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for nix-verification-loop

Your own site · 80×15
<a href="https://agentmods.dev/skills/filipnowakowicz/nixos-config/nix-verification-loop"><img src="https://agentmods.dev/badge/skills/filipnowakowicz/nixos-config/nix-verification-loop.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 29 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,487 The whole file, excluding the scripts and references it only reads on demand.
Security scan B 1 finding. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00029 $0.02487
Opus 5 $0.00015 $0.01243
Sonnet 5 $0.00006 $0.00497
Haiku 4.5 $0.00003 $0.00249

Measured 9d ago against content hash 70ecaa8bb01d, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade B, and why

nix-verification-loop scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Asks for rootmediumPrivilege escalation

A mod that escalates privileges can change anything on the machine, not only the project.

the new services absent from the running generation. Confirm sudo is usable
.agents/skills/nix-verification-loop/SKILL.md · 155 lines

How it starts

The opening of the file, as written. The whole thing — 155 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Nix Verification Loop

Use this workflow whenever editing Nix, shell scripts, host definitions, tests, packages, deploy wiring, secrets boundaries, or generated data in this repo.

Workflow

  1. Identify the touched surface:
    • Flake/module wiring: start with bash scripts/validate.sh flake-eval.
    • Shared modules/profiles/lib helpers: add bash scripts/validate.sh light.
    • A single host: run bash scripts/validate.sh host <name> when closure risk matters.
    • Package outputs: run bash scripts/validate.sh package <name> or package all.
    • Profile behavior: run bash scripts/validate.sh profile-test <name>.
    • Homeserver routing: run bash scripts/validate.sh smoke-homeserver-gcp when endpoint behavior changes.
    • Markdown-only repo docs: run bash scripts/validate.sh docs.
  2. If new files are created for a Nix flake check or build, stage them before expecting Nix to see them.
  3. Prefer the narrow check first, then widen only when the touched surface justifies it.
  4. Report any skipped validation explicitly with the reason.
  5. For a wider tier (hosts, package all, heavy, cve-reports), consider running it via a subagent that redirects output to a file and returns only the exit status and tail — this keeps the noisy build entirely out of the main session's transcript instead of relying on truncation alone.

Guardrails

  • Do not run deploy commands unless the user asked for deployment or the task explicitly requires landing live host state.
  • Treat deploy-rs silence in non-interactive sessions as a known behavior; if necessary, fall back to closure build/copy/switch instructions from CLAUDE.md.
  • Never edit encrypted secrets directly. Use sops.
  • nix develop -c <tool> / nix develop --command <tool> now pass through <tool>'s stdout and exit status: the default devShell shell hook only execs zsh for interactive sessions ($- containing i), not for -c invocations. Prefer tools already available in the current shell directly when possible, but nix develop -c <cmd> is valid proof a command ran.
  • When backgrounding a validation command, join every step on one logical line with &&/; (or a single quoted bash -c '...') — never rely on a literal newline between top-level statements. A backgrounded multi-line command can collapse onto one pipeline, e.g. turning nix fmt FILE | tail -5 + a following git add line into tail -5 git add FILE && git commit, which hangs tail forever on an open pipe with empty output. Avoid piping nix develop -c <cmd> through tail/head in the background for the same reason; run such commands in the foreground or redirect to a file instead.

Read the full file on GitHub · 155 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 9d ago First seen · 155 lines · 29 tokens per session scan B 70ecaa8bb01d

Subscribe to this mod's changes

nix-verification-loop is a skill published in the GitHub repository FilipNowakowicz/nixos-config (5 stars, last pushed 5d ago), licensed MIT. It adds 29 tokens to every session and 2,487 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it B with 1 finding (asks for root). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.