agentmods.dev

sql-benchmarks-dagster AGENTS.md

Instructions for running repeatable SQL performance experiments in a benchmarking laboratory, using files, an HTTP API, or MCP tools.

In plain words
What is it for?
They are for submitting benchmark configurations, reading result capsules, comparing database engines, and requesting engine recommendations.
Why use it?
They let an agent test a performance claim and verify the recorded result instead of relying on an untested assumption.

Instructions file for CodexOpenCode

2 1mo ago A scan 1,830 tokens original Apache-2.0
View the source file rctruta/sql-benchmarks-dagster
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.

agentmods
npx agentmods add instructions/rctruta/sql-benchmarks-dagster/agents-md
Clone the repo
git clone --depth 1 https://github.com/rctruta/sql-benchmarks-dagster

Made for: Codex, OpenCode.

Per session 1,830 This file is loaded in full into every session.
When invoked 1,830 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
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 $0.01830 $0.01830
Opus 5 $0.00915 $0.00915
Sonnet 5 $0.00366 $0.00366
Haiku 4.5 $0.00183 $0.00183

Measured 2d ago against content hash 7327623f4431, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

sql-benchmarks-dagster AGENTS.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 2d 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.

AGENTS.md · 150 lines

How it starts

The opening of the file, as written. The whole thing — 150 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agentic Benchmarking Protocol

Specification for AI agents and orchestrators interacting with the SQL Benchmarking Laboratory.

The lab is a deterministic performance oracle: every experiment is identified by a content-addressed hash, every result is a verifiable capsule, and every interface below is machine-first. An agent can verify a performance hypothesis here instead of asserting it.

Three ways in

  1. Filesystem + CLI (zero infrastructure): write a YAML config, run it, read the capsule.
  2. REST API (python serve.py, then http://localhost:8000/docs): catalog, results, cross-engine comparison, engine recommendation, async experiment submission.
  3. MCP server (python mcp_server.py): the same capabilities as native tools for MCP-capable agents. Requires the REST API to be running. (Note: This repository is natively Agent-Ready. It ships a claude.json configuration, meaning Claude Code will automatically mount the MCP server upon entering the directory.)

The agentic loop (CLI form)

  1. Hypothesis — e.g. "Quack's attach mode degrades with scan size."
  2. Submit — write a YAML config to sql_benchmarks/experiments/queue/ (configs outside the experiments directory are rejected). Start from a curated template — either the fully-annotated sql_benchmarks/experiments/templates/experiment_template.yaml, or an existing valid experiment in sql_benchmarks/experiments/queue/ (e.g. quickstart.yaml). REST-API agents fetch templates via GET /v1/catalog/templates and GET /v1/catalog/templates/{name} — see "Template discovery" below.
  3. Execute — ./run.sh sql_benchmarks/experiments/queue/my_exp.yaml --auto
  4. Identity — the system derives the 8-character Experiment ID: a SHA-256 fingerprint of the config + the SQL + all measurement-relevant Python (orchestration, engine clients, data generators). Formatting and comments never change an ID; logic always does.
  5. Retrieve — sql_benchmarks/experiments/results/<ID>/:
    • <ID>.csv — flattened matrix: one row per (engine × partition), with Duration, Duration_Min/Duration_Max (replication spread), and DNF (did-not-finish: an engine limitation recorded as data, not a crash)
    • fragments/*.json — atomic per-partition measurements including durations_raw (every replication, not just the mean)
    • metadata_<ID>.json — the conditions: engine/Python versions, OS, machine, cores, RAM
    • experiment_config.yaml — the exact config that ran, archived verbatim (the author's source bytes, not a re-serialization)
    • queries/<dialect>/*.sql — the exact SQL each engine ran (the dialect set the config's engines selected)

Read the full file on GitHub · 150 lines

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. 2d ago First seen · 150 lines · 1,830 tokens per session scan A 7327623f4431

Subscribe to this mod's changes

sql-benchmarks-dagster AGENTS.md is an instructions file published in the GitHub repository rctruta/sql-benchmarks-dagster (2 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 1,830 tokens to every session, about $0.0092 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.

Backend and APIsDatabasesDevOps and cloudbenchmarksdagsterdocker-composeduckdbpostgresqlsql
Related

Other instructions, from other repositories

graphjin AGENTS.md

Instructions for dosco/graphjin, covering graphjin agent guide, architectural overview, directory structure & responsibilities, build commands and coding guidelines.

dosco/graphjin · 5,837 tokens

evidence AGENTS.md

AGENTS.md instructions for evidence-dev/evidence, covering agents.md, this repo is a mirror, layout and scripts.

evidence-dev/evidence · 289 tokens

mssql-extension CLAUDE.md

Instructions for hugr-lab/mssql-extension, covering mssql-extension development guidelines, technology, project structure, versioning and commands.

hugr-lab/mssql-extension · 11,822 tokens

WrenAI AGENTS.md

AGENTS.md instructions for Canner/WrenAI: Repository-wide instructions live in .claude/CLAUDE.md. Please refer to that file for shared build commands, architecture, and conventions.

Canner/WrenAI · 123 tokens

sidemantic AGENTS.md

AGENTS.md instructions for sidequery/sidemantic, covering sidemantic, usage preference (important), codebase structure, version management and validation before commits.

sidequery/sidemantic · 1,674 tokens

sidemantic CLAUDE.md

Claude Code instructions for sidequery/sidemantic, a project described as: The universal metrics layer. Compatible with 15+ formats: Cube, MetricFlow, LookML, Omni, BSL, LDM, Cortex, Malloy, OSI, SML, TML, Hex, Rill, Superset.

sidequery/sidemantic · 6 tokens