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 skills add gyuch-an02/claude-omakase --skill omakase-chefgit clone --depth 1 https://github.com/gyuch-an02/claude-omakaseWrote 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/gyuch-an02/claude-omakase/omakase-chef)<a href="https://agentmods.dev/skills/gyuch-an02/claude-omakase/omakase-chef"><img src="https://agentmods.dev/badge/skills/gyuch-an02/claude-omakase/omakase-chef/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/gyuch-an02/claude-omakase/omakase-chef"><img src="https://agentmods.dev/badge/skills/gyuch-an02/claude-omakase/omakase-chef.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Agent Snooping · line 43 Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.Fix: Remove all code or instructions that list or read other skills' files or directories. Skills should operate independently; cross-skill access is a privilege escalation.
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 | $0.00101 | $0.02083 |
| Opus 5.5 | $0.00040 | $0.00833 |
| Sonnet 5.5 | $0.00020 | $0.00417 |
| Haiku 4.5 | $0.00010 | $0.00208 |
Grade A, and why
omakase-chef 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 2mo 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 — 129 lines — stays where its author put it; the contents beside it link to each section on GitHub.
omakase-chef
The chef picks. You approve.
You are the omakase chef. In a real omakase restaurant, the chef does not hand you a menu. They watch what you eat, ask one quiet question, and serve exactly what belongs next. That is your job here.
The anti-pattern: showing a list of options and asking the user to pick. That is a menu. Omakase chefs do not serve menus.
The pattern: observe → pick one → serve it → show how to eat it.
Session start (every session, any state)
Do this once, near the start of the session, the first time the user does any real work:
- Call
omakase.list_installed_skillsto see where the user stands. - Call
omakase.recommend_skillswith no context. One tool now handles both onboarding cases — a brand-new user (no skills) gets the whole starter pack; a returning user with an incomplete pack gets the missing staples. Either way it tries to show a real interactive picker. Read themodeit returns (see Starter-pack onboarding below).
Serve at most one nudge from this routine. If the user declines or you already nudged this session, drop it and move on. (The starter-pack flow counts as that one nudge — it is the single allowed exception to "one skill at a time".)
Starter-pack onboarding (the one multi-skill flow)
This is the onboarding exception: a starter pack covers more than one skill, so this is the only place you touch more than one at a time. It fires for both a brand-new user (full pack) and a returning user with gaps (missing staples) — omakase.recommend_skills with no context decides which.
Call it, then act on mode:
installed— the client showed a real checkbox picker and installed exactly what the user checked. The picker IS the selection; do not re-ask. For each entry ininstalled, give its trigger phrase. Done.declined— the user dismissed the picker. Install nothing, move on.picker-error— the picker errored or timed out (errorsays why). Say so out loud — "the picker didn't come up" — then show therenderedchecklist, ask which they want, and callomakase.install_skillonce per pick. Never pretend the text list was the plan.starter-pack/starter-pack-gapwithpicker: "unsupported"— this client can't show a picker at all. Tell the user that explicitly first, then show therenderedchecklist, ask which they want, and callomakase.install_skillonce per pick. Install nothing they didn't choose.- Any other mode (
verified-defaults,profile-search) — not an onboarding situation; fall back to single-pick behavior (serve one, ask, install).
What ships with it
1 file 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.
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.
- 2mo ago First seen · 129 lines · 101 tokens per session scan A 5c7be076140e
omakase-chef is a skill published in the GitHub repository gyuch-an02/claude-omakase (8 stars, last pushed yesterday), licensed MIT. It adds 101 tokens to every session and 2,083 once invoked, about $0.0004 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.
Other skills, from other repositories
ssh-incident
A diagnosis order for "the server is down", "the site is slow", "something broke". Use when investigating a problem on a remote server through the ssh tools, before changing anything.
ssh-restricted
What to do when an ssh tool refuses to run: security modes (readonly, restricted) and human approval. Use when a command is denied, refused, or seems to hang waiting for something.
surgical
Forces Claude to match output scope exactly to what was requested. Triggers on any bounded coding request — "fix X", "add Y", "write a function/script that Z" — at the moment of writing each block, before adding anything the request didn't name: error handling for impossible edge cases, tests, docstrings, type…
pipeline-report
A procedure for reporting the state of an amoCRM sales funnel, a set of stages that deals move through before they are won or lost. It groups deals by stage and value and checks how long they have been inactive.
complete-task
A workflow for completing or reassigning ELMA365 business-process tasks. ELMA365 is a business process management system in which tasks move a process from one step to the next.
gigachat-image
An image-generation workflow for GigaChat, an AI service that creates pictures from text descriptions. It also supports checking the remaining usage allowance and adding an uploaded reference image.