agentmods.dev

sy-automl-mcp: Instructions file for Claude Code

CLAUDE.md

sy-automl-mcp CLAUDE.md is an instructions file for Claude Code from noahwang550/sy-automl-mcp. It costs 6,361 tokens per session, scanned A, original, Apache-2.0.

Repository instructions for sy-automl-mcp, a server for automated machine-learning tasks through MCP. They describe its completed work, technology choices, and project constraints.

In plain words
What is it for?
Understanding the project while developing, testing, linting, or maintaining its tabular, time-series, and multimodal machine-learning workflows.
Why use it?
They give coding agents the project context needed to change the server without overlooking its status or required safeguards.

Instructions file for Claude Code

Written for Claude Code: the file is CLAUDE.md. Also seen: mentions CLAUDE.md; mentions Claude Code.

This is noahwang550/sy-automl-mcp's own configuration. It tells Claude Code how to work on sy-automl-mcp itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything sy-automl-mcp configures →

not rated 0repo changed 1mo ago A scan 6,361 tokens original Apache-2.0
View source ↗ noahwang550/sy-automl-mcp
Reuse

Borrowing it

Nothing to install: this file belongs to noahwang550/sy-automl-mcp. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/noahwang550/sy-automl-mcp/master/CLAUDE.md
Clone the repo
git clone --depth 1 https://github.com/noahwang550/sy-automl-mcp

Made for: Claude Code.

Wrote 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.

agentmods badge for sy-automl-mcp CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/noahwang550/sy-automl-mcp/claude-md/github.svg)](https://agentmods.dev/instructions/noahwang550/sy-automl-mcp/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/noahwang550/sy-automl-mcp/claude-md"><img src="https://agentmods.dev/badge/instructions/noahwang550/sy-automl-mcp/claude-md/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.

agentmods 80×15 button for sy-automl-mcp CLAUDE.md

Your own site · 80×15
<a href="https://agentmods.dev/instructions/noahwang550/sy-automl-mcp/claude-md"><img src="https://agentmods.dev/badge/instructions/noahwang550/sy-automl-mcp/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 6,361 This file is loaded in full into every session.
When invoked 6,361 The same file — it is already loaded in full.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.06361 $0.06361
Opus 5.5 $0.02544 $0.02544
Sonnet 5.5 $0.01272 $0.01272
Haiku 4.5 $0.00636 $0.00636

Measured 1mo ago against content hash 4753c06316d5, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

sy-automl-mcp 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 1mo 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.

CLAUDE.md · 156 lines

How it starts

The opening of the file, as written. The whole thing — 156 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Project Purpose

Wrap AutoGluon (AWS's open-source AutoML framework) into an MCP (Model Context Protocol) server so that AI assistants such as Claude can drive AutoML workflows — loading datasets, training models, predicting, and evaluating — through standard MCP tool calls.

AutoGluon capabilities exposed:

  • TabularPredictor — classification/regression on structured tables (fit, predict, evaluate, leaderboard, feature_importance, fit_summary).
  • TimeSeriesPredictor — forecasting (target, prediction_length, freq).
  • MultimodalPredictor — image, text, and multimodal tasks.

Tech Stack

  • Language: Python 3.11 (AutoGluon 1.6.1 verified).
  • MCP server: mcp Python SDK, FastMCP decorator style (@mcp.tool()). Transports: stdio (default) and streamable-http.
  • AutoGluon: autogluon.tabular 1.6.1, autogluon.timeseries 1.6.1, autogluon.multimodal 1.6.1.
  • Docker: python:3.11-slim base, tiered build (tabular vs full). pandas 2.3.3.
  • Word report (v0.7.3): python-docx 1.1.2 for business-facing report.docx generation; lazily imported inside tools/report_docx.py so tabular-tier images without it still train successfully (docx is best-effort, failure-isolated).

Critical Constraints

  • Docker is the primary runtime, not native host Python. AutoGluon officially supports Linux/macOS only; this project runs the MCP server inside a Docker container regardless of host OS.
    • Dockerfile based on python:3.11-slim, tiered install via --build-arg TIER=tabular|full.
    • artifacts/ bind-mounted at /app/artifacts for persistence.
    • Image ENTRYPOINT is python server.py — override with --entrypoint sh or --entrypoint python for pytest.
    • pytest is NOT in the production image — install at runtime (pip install pytest pytest-asyncio -q) before running tests.
    • Tabular tier runs on CPU; full tier benefits from --gpus all.
  • Training is long-running. fit() can take minutes to hours. Background-task pattern: train_* returns task_id, poll via get_task_status / get_task_result.
  • Heavy dependencies. AutoGluon + torch pull several GB. Docker isolates this.
  • Stdout must stay clean. AutoGluon/PyTorch/Lightning write progress bars + banners to stdout/stderr, which would corrupt MCP stdio JSON-RPC. See Stdout Pollution Fix below.
  • Env vars:
    • MCP_TRANSPORT (stdio default | http for streamable-http)
    • MCP_HOST / MCP_PORT (default 0.0.0.0 / 8000)
    • MCP_API_TOKEN (default: unset) — Bearer token for streamable-http auth. Unset/empty = auth disabled (backward-compatible, fine for trusted/local use).
    • MCP_MODEL_CACHE_MAX (default 4) — LRU cap for in-memory predictor cache.
    • MCP_TASK_RETENTION_SECONDS (default 86400) — TTL for terminal task records before sweep.
    • MCP_TASK_MAX_RETAINED (default 100) — max terminal tasks retained before oldest-first sweep.
    • MCP_MAX_WORKERS (default 1) — background-task thread pool size. Safe to raise above 1 because stdout/stderr redirection is thread-local (see Stdout section).
    • MAX_UPLOAD_CHUNK_BYTES (default 1048576) — max decoded size of a single upload_dataset_chunk payload.
    • MAX_UPLOAD_TOTAL_BYTES (default 268435456) — max assembled size of a finalize_dataset output.
    • MCP_DOWNLOAD_ENABLED (default true) — enables the /download artifact-byte bridge (presigned HMAC URLs, Range support). Fail-closed.
    • MAX_DOWNLOAD_BYTES (default 1073741824) — per-request byte cap on /download (1GB). Range segments also capped.
    • MCP_ARTIFACT_BASE_URL (default: unset) — external base URL used by get_artifact_url to build presigned download links (e.g. https://host:9885). Required for the download bridge to work end-to-end.
    • MCP_DOWNLOAD_SIGNING_KEY (default: unset) — HMAC-SHA256 signing key for presigned download URLs. Required (fail-closed: /download returns 503 and get_artifact_url returns failure envelope when unset).
    • MCP_DOWNLOAD_URL_TTL_SECONDS (default 3600) — TTL for presigned download URLs (clamped to [60, 86400]).
    • MAX_INLINE_BYTES (default 8388608 / 8MB, hard-capped at MAX_DOWNLOAD_BYTES) — byte cap for get_artifact_bytes, which returns file content base64-encoded directly in the MCP tool response. Use when the agent platform's security layer auto-redacts token query strings (which breaks presigned get_artifact_url links). Larger files must go through get_artifact_url.
    • MAX_CHUNK_BYTES (default 65536 / 64KB, hard-capped at MAX_INLINE_BYTES) — per-call byte cap for get_artifact_chunk, the chunked-pull fallback for large files that exceed the platform's tool-response redaction threshold. Agent loops offset=0, length, 2*length, ... until data.final=true, concatenating content_base64 slices in order.

Read the full file on GitHub · 156 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. 1mo ago Changed · +8 lines · +1,356 tokens per session 4753c06316d5
  2. 1mo ago Changed · -116 lines · -2,078 tokens per session 39c1b1988e17
  3. 1mo ago First seen · 264 lines · 7,083 tokens per session scan A bb80a4166c1c

Subscribe to this mod's changes

sy-automl-mcp CLAUDE.md is an instructions file published in the GitHub repository noahwang550/sy-automl-mcp (0 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 6,361 tokens to every session, about $0.0254 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 2026-09-01.

Building agents
Related

Other instructions, from other repositories

vscode buildNext.instructions.md

Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).

microsoft/vscode · 9,715 tokens

next.js AGENTS.md

AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.

vercel/next.js · 7,581 tokens

vscode oss-third-party-notices.instructions.md

Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, cache eligibility and rollout and applying the notice (cutover).

microsoft/vscode · 5,262 tokens

langchain AGENTS.md

AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.

langchain-ai/langchain · 4,680 tokens

vscode copilot-instructions.md

Copilot instructions for microsoft/vscode, covering vs code copilot instructions, project overview, root folders, core architecture (src/ folder) and built-in extensions (extensions/ folder).

microsoft/vscode · 2,731 tokens

deno copilot-instructions.md

Copilot instructions for denoland/deno, covering deno development guide for github copilot, network access, high level overview, key directories and key files to understand first.

denoland/deno · 3,077 tokens