Borrowing it
Nothing to install: this file belongs to tpierrain/kenjaku. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/tpierrain/kenjaku/main/.claude/skills/test-first-discipline/SKILL.mdgit clone --depth 1 https://github.com/tpierrain/kenjakuWrote 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.
[](https://agentmods.dev/skills/tpierrain/kenjaku/test-first-discipline)<a href="https://agentmods.dev/skills/tpierrain/kenjaku/test-first-discipline"><img src="https://agentmods.dev/badge/skills/tpierrain/kenjaku/test-first-discipline/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.
<a href="https://agentmods.dev/skills/tpierrain/kenjaku/test-first-discipline"><img src="https://agentmods.dev/badge/skills/tpierrain/kenjaku/test-first-discipline.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00117 | $0.06096 |
| Opus 5 | $0.00059 | $0.03048 |
| Sonnet 5 | $0.00023 | $0.01219 |
| Haiku 4.5 | $0.00012 | $0.00610 |
Grade A, and why
test-first-discipline 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 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.
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.
How it starts
The opening of the file, as written. The whole thing — 351 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Test-first discipline (universal)
The name states what SURVIVED the measurement, not what was dropped: the test comes before the code, and it is seen red for the right reason first. Everything else in here is a mode, a reflex or a judge.
This skill replaces
tdd-discipline(v1.1.0), and the replacement is not cosmetic. That skill opened on a rule — "one test at a time, complete red→green→refactor at each step; writing several tests ahead then implementing is forbidden" — which was measured and did not survive. Everything else it carried (fail-first, the refactor step, and above all the assertion-quality reflexes below) was kept verbatim, because that is what the measurements credit.Why the name changed rather than the clause. What was dropped is not a ceremony inside TDD, it is TDD's own thesis: the design emerging under the pressure of one example at a time. What remains — specify, see it fail, implement, refactor, and let a mutation score judge — is honest work, but calling it TDD would be false advertising. Classic TDD did not disappear; it became a mode, described below with its criteria and its open question.
The evidence (Kenjaku,
maintainers/mutation/RESULTS.md), first-pass scores on brand-new files, before any review and before any survivor was killed:
Release Mode New files, first pass v4.9.0 strict baby-steps 84.62 % and 87.74 % v4.9.1 design-first, test-first in small batches 91.80 % One release is one data point, and the v4.9.1 figure only reached 95.92–100 % after three adversarial reviews and two more mutation passes. So the claim is not "the relaxed mode is free". It is: the step size did not carry the quality — the nets did.
🛑 The low numbers you are remembering are NOT baby-steps — here is what each one is
No score in this record is low because the steps were small. Every low number that gets quoted against baby-steps belongs to a different cause, and the cause is always when the test was written or whether the line sits in a seam a test can reach:
Recalled as "baby-steps scored badly" Actually What it really measures 51.5 % — update-engine.mjscomposition glue written after an already-green core test-after, never step size 71.4 % — reconcile-brain.mjsa pure core wrapped in I/O and a CLI an unreachable branch (reflex 6) 66 %, and a flat 0 % for months top-level entry points no test ever imported the entry-point seam rule And the reverse case, from the same audit:
engine-skill-refresh.mjs— the one file built in strict baby-steps — came out top of its batch at 86.7 %, needing "only assertion polish". The two numbers actually attached to baby-steps in this record are 84.62 % and 87.74 %. There is no ~60 % baby-steps run. It does not exist.Why this table is here rather than left to be re-derived: the misattribution has now happened twice, from both sides (the 87-vs-51 pair was long quoted in this file's own predecessor as evidence about step size; the owner recalled a "60 % baby-steps" figure on 2026-08-20). A number misfiled twice is a number that will be misfiled again — so the filing is written down, not the correction.
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.
- 9d ago First seen · 351 lines · 117 tokens per session scan A cde29d8c6bcf
test-first-discipline is a skill published in the GitHub repository tpierrain/kenjaku (80 stars, last pushed today), licensed Apache-2.0. It adds 117 tokens to every session and 6,096 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.
Other skills, from other repositories
brainstorming-and-tdd
Clarify behavior and use test-first implementation when applicable.
lint
A health check for the LLM Wiki, a knowledge base about large language models. It looks for missing pages, broken links, conflicting or outdated content, missing fields, incomplete topic maps, weak bias checks, and an old core-context snapshot.
test-driven-development
Use when implementing any feature or bugfix, before writing implementation code.
engram-testing-coverage
TDD and coverage standards for Engram. Trigger: When implementing behavior changes in any package.
iterative-development
TDD iteration loops using Claude Code Stop hooks - runs tests after each response, feeds failures back automatically.
python
Python development with ruff, mypy, pytest - TDD and type safety.