field-test

field-test is a skill for Claude Code, Codex from cloudposse/atmos. It costs 116 tokens per session (4,320 once invoked), scanned A, original, Apache-2.0.

A manual, adversarial test process for a feature or command-line command. It combines reading the implementation and tests with trying realistic user mistakes and misuse that automated tests may not cover.

In plain words
What is it for?
Use it to test a feature or CLI command against real state, create reusable test fixtures, investigate plausible failure cases, and produce a ranked report of findings.
Why use it?
It finds usability and behavior problems that ordinary tests can miss, including misunderstandings, bad inputs, and failures involving real state. The process reports findings but does not change the code.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/cloudposse/atmos/field-test
Any agent
npx skills add cloudposse/atmos --skill field-test
Clone the repo
git clone --depth 1 https://github.com/cloudposse/atmos

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 field-test

README.md
[![agentmods](https://agentmods.dev/badge/skills/cloudposse/atmos/field-test.svg)](https://agentmods.dev/skills/cloudposse/atmos/field-test)
Your own site
<a href="https://agentmods.dev/skills/cloudposse/atmos/field-test"><img src="https://agentmods.dev/badge/skills/cloudposse/atmos/field-test.svg" alt="Measured on agentmods" height="20"></a>
Per session 116 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 4,320 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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 $0.00116 $0.04320
Opus 5 $0.00058 $0.02160
Sonnet 5 $0.00023 $0.00864
Haiku 4.5 $0.00012 $0.00432

Measured 4d ago against content hash 17d300cccf14, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

field-test scanned grade A with 0 findings 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 4d 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.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/skills/field-test/SKILL.md · 249 lines

How it starts

The opening of the file, as written. The whole thing — 249 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Field Test

Hands-on, adversarial test pass of $ARGUMENTS (the feature/command named when this skill was invoked, e.g. atmos vendor pull).

If no target was given, default to the change introduced on the current branch rather than asking. Resolve the actual pull-request base branch when one is available (gh pr view --json baseRefName -q .baseRefName for the current branch), falling back to the repository's default branch (gh repo view --json defaultBranchRef -q .defaultBranchRef.name) when no PR exists yet. Do NOT use the upstream tracking branch (@{u}) as the base — for a normal feature branch that tracks origin/<same-branch-name>, diffing against its own upstream produces an empty or near-empty diff, not the PR's actual changes, once the branch has been pushed.

A branch NAME from gh (e.g. develop) is not guaranteed to be a usable git ref in THIS checkout — shallow clones, detached HEADs, and worktrees with a narrow fetch refspec can have the name without the commits. Before running any diff, verify the base actually resolves here (git rev-parse --verify --quiet <candidate>^{commit}), trying in order: origin/<resolved-name>, <resolved-name> — take the first that verifies. Do not fall back to origin/main/main when resolved-name came from an actual PR base other than the default branch (e.g. a PR targeting develop) — diffing against main instead of the PR's real base compares against the wrong history and can silently miss real changes or include unrelated ones. origin/main/main are legitimate fallback candidates only in the no-PR case (gh pr view returned nothing, so resolved-name already IS the repository's default branch) or when gh itself isn't available at all. If a known, non-default PR base doesn't resolve as either origin/<resolved-name> or <resolved-name>, stop and ask the user which base to diff against — do not run `git diff

Goal: catch "vibe-coded slop" — behavior that looks fine in code review but breaks or misleads a real user — not to re-run what automated tests already cover. Anticipate plausible user misunderstandings, not just obvious bugs.

This pass is investigation only. Do not fix anything you find — see Phase 5. Report and stop; the user decides what to fix. Any follow-up fix work — including a plan to fix the findings, not just the implementation — must close with the fix-log skill, not this one.

Read the full file on GitHub · 249 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. 4d ago First seen · 249 lines · 116 tokens per session scan A 17d300cccf14

Subscribe to this mod's changes

field-test is a skill published in the GitHub repository cloudposse/atmos (1,369 stars, last pushed today), licensed Apache-2.0. It adds 116 tokens to every session and 4,320 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.