DB Benchmark

What it does

When your agent uses it

• A new repository read function is added
• Benchmark cases are missing
• Working on scripts/benchmark-db

Example prompts

• “/db-benchmark”

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

1. 1Add a field to the Fixtures interface. Every field is T | null — null means "table empty, skip dependent cases". Bundle related values…
2. 2Add a resolve() helper at the bottom of the file and call it from resolveFixtures().
3. 3Pick the heaviest row, not any row — the standard pattern is a groupBy + count + orderBy count desc + limit 1 query
4. 4Never throw on empty tables — always resolve to null (executeTakeFirst() + ?? null).
5. 5Verify table/column names against app/db/tables.ts.

What it can do on your machine

Read from SKILL.md and the folder at commit 63d970e. It shows what the files ask for, not the result of running them.

• Tool permissions

  Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

  From allowed-tools in the SKILL.md frontmatter.

• Runs code

  Shell commands in SKILL.md call:

  • pnpm

  From the folder's file list and the shell code blocks in SKILL.md.

• Network

  No URLs in SKILL.md. Its commands use pnpm, which can reach the network depending on how they are called.

  From URLs in SKILL.md, links to its own repository left out.

• Credentials

  Names no API keys, tokens, secrets or passwords.

  From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

DB Benchmark loads about 1.1k tokens when it runs. Until then it costs about 58 tokens; SKILL.md has 378 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it

~58

When it runs · the whole SKILL.md, loaded when a task matches

~1.1k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from sendou-ink/sendou.ink at commit 63d970e, republished under its AGPL-3.0 licence (© sendou-ink). 378 words, ~1,052 tokens.

name

db-benchmark

description

Add or update repository READ function benchmark cases in the DB benchmark script. Use when a new repository read function is added, when benchmark cases are missing or outdated, or when working on scripts/benchmark-db.

// Function needs a fixture (skipped automatically if the fixture is null)
add("UserRepository.findLeanById", fx.heavyUser, (user) =>
	UserRepository.findLeanById(user.id),
);

// Function takes no arguments
addStatic("BadgeRepository.all", () => BadgeRepository.all());

// Function needs two fixtures
add(
	"SavedCalendarEventRepository.isSaved",
	both(fx.heavyUser, fx.heavyTournamentId),
	([user, tournamentId]) =>
		SavedCalendarEventRepository.isSaved({ userId: user.id, tournamentId }),
);

async function resolveHeavyBuildUserId() {
	const row = await db
		.selectFrom("Build")
		.select(({ fn }) => ["ownerId", fn.countAll<number>().as("count")])
		.groupBy("ownerId")
		.orderBy("count", "desc")
		.limit(1)
		.executeTakeFirst();

	return row?.ownerId ?? null;
}

pnpm run typecheck:scripts
pnpm exec biome check --error-on-warnings --write scripts/benchmark-db.ts scripts/benchmark-db
DB_PATH=db-prod.sqlite3 pnpm exec vite-node scripts/benchmark-db.ts -- --filter <caseName> --iterations 3

© sendou-ink, AGPL-3.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .claude/skills/db-benchmark of sendou-ink/sendou.ink.

Open the folder on GitHubat commit 63d970e