self-graded-green

self-graded-green is a skill for Claude Code, Codex from arunveersingh/ai. It costs 0 tokens per session (533 once invoked), scanned A, original, Apache-2.0.

A development practice that makes an agent turn each requirement into concrete input-and-expected-result examples before writing code. After approval, those examples become tests; TDD means designing tests before or alongside the implementation.

In plain words
What is it for?
Use it to review requirements, identify alternate interpretations, and get approval before implementation. Then use the approved examples as tests for the code.
Why use it?
It reduces the risk that the agent silently chooses the wrong meaning for an ambiguous requirement. It also exposes mistakes such as checking each refund separately when the rule concerns the total.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Good fit Use it to review requirements, identify alternate interpretations, and get approval before implementation. Then use the approved examples as tests for the code.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/arunveersingh/ai/day-08-self-graded-green
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.

Any agent
npx skills add arunveersingh/ai --skill day-08-self-graded-green
Clone the repo
git clone --depth 1 https://github.com/arunveersingh/ai

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 self-graded-green

README.md
[![agentmods](https://agentmods.dev/badge/skills/arunveersingh/ai/day-08-self-graded-green/github.svg)](https://agentmods.dev/skills/arunveersingh/ai/day-08-self-graded-green)
Your own site
<a href="https://agentmods.dev/skills/arunveersingh/ai/day-08-self-graded-green"><img src="https://agentmods.dev/badge/skills/arunveersingh/ai/day-08-self-graded-green/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 self-graded-green

Your own site · 80×15
<a href="https://agentmods.dev/skills/arunveersingh/ai/day-08-self-graded-green"><img src="https://agentmods.dev/badge/skills/arunveersingh/ai/day-08-self-graded-green.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 0 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 533 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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00000 $0.00533
Opus 5.5 $0.00000 $0.00213
Sonnet 5.5 $0.00000 $0.00107
Haiku 4.5 $0.00000 $0.00053

Measured 15d ago against content hash 30ac1c965155, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

self-graded-green 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 15d 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.

courses/make-ai-fail-safely/practice/day-08-self-graded-green/SKILL.md · 38 lines

What it actually says

Self-graded green

The requirement: "Refunds must not exceed the amount paid." The agent reads "each refund," writes the check and a test (20000 on 10000: rejected), and everything passes. You meant the total. Two refunds of 6000 on a 10000 payment both go through.

Self-graded is not passed. Green means the agent agrees with itself. The fix: see its reading as examples before it builds, and approve that first.

Part A — Ask (paste and adapt)

Before writing any code:
1. For each requirement, give 2-4 concrete examples:
   input -> expected result. Include repeats, totals,
   boundaries, and retries where they apply.
2. If a requirement reads two ways, write TWO READINGS:
   <requirement>, show an example of each, and don't pick.
3. Stop and wait for my approval.
After I approve: build to those examples, turn them into
tests, and don't change them. If one looks wrong, ask.

In an orchestrated flow (spec → plan → tasks → implement): run this as one step right after the spec. On a big requirement, run it per slice, just before that slice's tasks.

Part B — Read the examples (you do this)

  1. Write your own first for one line you care about, then compare. A mismatch is a misreading caught for free.
  2. Look for the split. For each requirement ask: would the other reading give the same result here? If yes for every example, add one where it wouldn't (two refunds, not one).
  3. Answer every TWO READINGS in the spec itself, so the next reader doesn't have to pick.
  4. Hold the line. Later tasks build to the approved examples. If the agent wants to change one, that's a spec question for you, not an edit.

Done when: every requirement has examples you read, at least one per line where the readings would differ, and no open TWO READINGS.

Lesson: Day 8 — Self-Graded Green

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. 15d ago First seen · 38 lines · 0 tokens per session scan A 30ac1c965155

Subscribe to this mod's changes

self-graded-green is a skill published in the GitHub repository arunveersingh/ai (3 stars, last pushed 3d ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 533 tokens. 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 2026-10-11.