What is floom-skill-evals?

A system for testing whether an agent skill improves results by comparing runs with and without that skill. TDD means test-driven development; here, the system instead uses isolated task workspaces and automatic checks to grade the runs.

Use it to create evaluation tasks, run baseline-versus-skill tests in Docker, check outputs with a verifier, and save reports.

It replaces vague claims about a skill being useful with repeatable comparisons and saved results.

floom-skill-evals — Skill for Claude Code and Codex by floomhq/starter

Install

Getting it into your agent

It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.

Clone the repo

git clone --depth 1 https://github.com/floomhq/starter

agentmods

npx agentmods add skills/floomhq/starter/floom-skill-evals

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.

README.md

[![agentmods](https://agentmods.dev/badge/skills/floomhq/starter/floom-skill-evals/github.svg)](https://agentmods.dev/skills/floomhq/starter/floom-skill-evals)

Your own site

<a href="https://agentmods.dev/skills/floomhq/starter/floom-skill-evals"><img src="https://agentmods.dev/badge/skills/floomhq/starter/floom-skill-evals/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.

Your own site · 80×15

<a href="https://agentmods.dev/skills/floomhq/starter/floom-skill-evals"><img src="https://agentmods.dev/badge/skills/floomhq/starter/floom-skill-evals.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 57 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 620 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. 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.

Model	Per session	Once invoked
Fable 5.1	$0.00057	$0.00620
Opus 5.5	$0.00023	$0.00248
Sonnet 5.5	$0.00011	$0.00124
Haiku 4.5	$0.00006	$0.00062

Measured 1mo ago against content hash 2a50452a1322, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

floom-skill-evals 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 1mo ago.

The scan reads SKILL.md. This mod also ships 6 executable files (references/evals/seo-internal-link-architecture/verifier.py, references/evals/seo-meta-canonical-robots/verifier.py, references/evals/seo-structured-data-validation/verifier.py, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

skills/floom-skill-evals/SKILL.md · 91 lines

How it starts

The opening of the file, as written. The whole thing — 91 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Floom Skill Evals

Run simple, deterministic skill receipts:

Create eval tasks for a skill.
Run baseline and with-skill conditions in isolated Docker workspaces.
Grade with deterministic verifier.
Save result.json and a markdown receipt.

When to use

Use this skill for requests like:

"evaluate this skill"
"baseline vs with skill"
"run skill receipts"
"prove skill lift"
"Floom Verified style run"

Commands

1) Scaffold an eval task

python3 scripts/scaffold_eval_task.py \
  --skill seo-audit \
  --task technical-homepage-audit \
  --out ./skills/seo-audit/evals

This creates:

<out>/<task>/task.md
<out>/<task>/input/
<out>/<task>/expected.json
<out>/<task>/verifier.py
<out>/<task>/eval.json

2) Run A/B eval

python3 scripts/run_skill_eval.py \
  --eval-dir ./skills/seo-audit/evals/technical-homepage-audit \
  --skill-path ~/.codex/skills/seo-audit \
  --agent codex

Outputs:

runs/<timestamp>/result.json
runs/<timestamp>/report.md
condition workspaces and logs

Default policy rejects scripted output commands (echo, printf, heredoc writes) in agent_command. Use --allow-scripted only for runner plumbing tests.

Execution model

baseline: run agent command without skill mount.
with_skill: run same command with skill mounted at /skill and env FLOOM_SKILL_PATH=/skill.
verifier: runs deterministic checker inside each condition workspace.
audit: simple deterministic summary from pass/fail/runtime deltas.

Required eval contract

eval.json must include:

id
skill
timeout_minutes
trials
agent
agent_command.baseline
agent_command.with_skill
verifier
expected_output

Command placeholders supported:

{task_dir} mounted as /task
{output_path} path in workspace where agent writes result

Notes

This v0 runner is Docker-first and local-first.
Host keeps credentials; container only gets mounted task workspace.
If API keys are needed, pass scoped env explicitly with --pass-env OPENAI_API_KEY etc.
Receipts include provenance hashes for verifier and both condition commands.

Read the full file on GitHub · 91 lines

Files

What ships with it

31 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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.

1mo ago First seen · 91 lines · 57 tokens per session scan A 2a50452a1322

Subscribe to this mod's changes

floom-skill-evals is a skill published in the GitHub repository floomhq/starter (3 stars, last pushed 5mo ago), licensed MIT. It adds 57 tokens to every session and 620 once invoked, about $0.0002 per session on Opus 5.5. 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 2026-08-31.

Building agents Claude Code agent-skills ai-agents anthropic codex-cli cursor

parallel-feature-development

Coordinate parallel feature development with file ownership strategies, conflict avoidance rules, and integration patterns for multi-agent implementation. Use this skill when decomposing a large feature into independent work streams, when two or more agents need to implement different layers of the same system…

wshobson/agents · 105 tokens

nft-standards

Implement NFT standards (ERC-721, ERC-1155) with proper metadata handling, minting strategies, and marketplace integration. Use when creating NFT contracts, building NFT marketplaces, or implementing digital asset systems.

wshobson/agents · 48 tokens

cost-optimization

Optimize cloud costs across AWS, Azure, GCP, and OCI through resource rightsizing, tagging strategies, reserved instances, and spending analysis. Use when reducing cloud expenses, analyzing infrastructure costs, or implementing cost governance policies.