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.
npx agentmods add instructions/2389-research/test-kitchen/claude-mdgit clone --depth 1 https://github.com/2389-research/test-kitchenWrote 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/instructions/2389-research/test-kitchen/claude-md)<a href="https://agentmods.dev/instructions/2389-research/test-kitchen/claude-md"><img src="https://agentmods.dev/badge/instructions/2389-research/test-kitchen/claude-md.svg" alt="Measured on agentmods" height="20"></a>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 | $0.01988 | $0.01988 |
| Opus 5 | $0.00994 | $0.00994 |
| Sonnet 5 | $0.00398 | $0.00398 |
| Haiku 4.5 | $0.00199 | $0.00199 |
Grade A, and why
test-kitchen CLAUDE.md 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 5d 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 — 239 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Test Kitchen Plugin
Overview
Test Kitchen provides two gate skills for parallel implementation:
| Skill | Gate | When |
|---|---|---|
test-kitchen:omakase-off |
Entry | FIRST on any build/create/implement request |
test-kitchen:cookoff |
Exit | At design→implementation transition |
Skills Included
| Skill | Triggers | Description |
|---|---|---|
test-kitchen:omakase-off |
(1) FIRST on build/create, (2) During brainstorming on indecision, (3) Explicit request | Wraps brainstorming, offers parallel design exploration |
test-kitchen:cookoff |
At "let's implement" moments | Wraps implementation, offers parallel execution |
Flow
"Build X" / "Create Y" / "Implement Z"
↓
┌─────────────────────────────────────┐
│ OMAKASE-OFF (entry gate) │
│ Wraps brainstorming │
│ │
│ Choice: │
│ 1. Brainstorm together │
│ 2. Omakase (3-5 parallel designs) │
└─────────────────────────────────────┘
↓
[Brainstorming / Design phase]
↓
Design complete, "let's implement"
↓
┌─────────────────────────────────────┐
│ COOKOFF (exit gate) │
│ Wraps implementation │
│ │
│ Choice: │
│ 1. Cookoff (2-5 parallel agents) │
│ 2. Single subagent │
│ 3. Local implementation │
└─────────────────────────────────────┘
↓
[Implementation]
Key Design Principle
Skills need aggressive triggers to work.
Skills can't passively detect "uncertainty" or "readiness" - they must claim specific moments in the conversation flow:
- Omakase-off: Claims the BUILD/CREATE moment (before brainstorming)
- Cookoff: Claims the IMPLEMENT moment (after design)
Both skills present choices to the user, allowing them to opt into parallel execution or continue with standard workflows.
Omakase-off (Entry Gate)
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.
- 5d ago First seen · 239 lines · 1,988 tokens per session scan A 8d9e6a252e0a
test-kitchen CLAUDE.md is an instructions file published in the GitHub repository 2389-research/test-kitchen (4 stars, last pushed 2mo ago), licensed MIT. It adds 1,988 tokens to every session, about $0.0099 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 2026-08-31.
Other instructions, from other repositories
jean AGENTS.md
AGENTS.md instructions for coollabsio/jean, covering ai agents, current status, core rules, codex app server schema and new sessions.
jean GEMINI.md
Gemini CLI instructions for coollabsio/jean, a project described as: A dev environment for AI agents.
blazediff AGENTS.md
Instructions for teimurjan/blazediff, covering blazediff agent rules, hard rules, quick commands and pointers.
insyra AGENTS.md
AGENTS.md instructions for HazelnutParadise/insyra, covering agents.md, what this repo is, required artifacts, change workflow — openspec (required) and acceleration (accel) operating contract.
insyra CLAUDE.md
Claude Code instructions for HazelnutParadise/insyra: Read AGENTS.md before doing any project work. Treat it as the project operating contract.
svaba CLAUDE.md
Claude Code instructions for walaj/svaba, covering claude.md — svaba working notes, project at a glance, versioning — claude owns this, build system and statistical model — the files that matter.