MODEL_CONSOLIDATION_PLAN

A technical plan for reorganising the NodeBench AI agent system. It covers which AI models to keep, how to structure directories, which SDK connections to replace, Python MCP servers, and five agent patterns.

In plain words
What is it for?
Use it to guide model consolidation, directory changes, SDK replacement, Python MCP integration, and implementation of the listed agent patterns.
Why use it?
It gives developers one reference for a broad system restructure, reducing uncertainty about approved models, project layout, integrations, and orchestration choices.

Agent

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 agents/homenshum/nodebenchai/model_consolidation_plan
Clone the repo
git clone --depth 1 https://github.com/HomenShum/NodeBenchAI
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 13,587 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. 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 $0.00000 $0.13587
Opus 5 $0.00000 $0.06793
Sonnet 5 $0.00000 $0.02717
Haiku 4.5 $0.00000 $0.01359

Measured 2d ago against content hash 88e9759c32bd, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

MODEL_CONSOLIDATION_PLAN scanned grade A with 1 finding 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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

const response = await fetch(config.baseUrl, {
backend/convex/domains/agents/MODEL_CONSOLIDATION_PLAN.md · 1,419 lines

How it starts

The opening of the file, as written. The whole thing — 1,419 lines — stays where its author put it; the contents beside it link to each section on GitHub.

2025 Agent Architecture Restructure Plan

Overview

This document outlines the complete restructure of the NodeBench AI agent system:

  1. Model Consolidation - Use ONLY approved 2025 models
  2. Directory Restructure - Clean separation of concerns
  3. Native SDK Removal - Replace with AI SDK providers
  4. Python MCP Integration - Convex-aware Python servers
  5. 5 Agent Patterns - Implement 2025 orchestration patterns

Approved Models (EXCLUSIVE - No Others Allowed)

  • OpenAI: GPT-5.2 series only
  • Anthropic: Claude 4.5 series only
  • Google: Gemini 3 and 2.5 series only

Part 1: Model Consolidation

Models to KEEP (7 Total)

Provider Model ID (Your Format) AI SDK Model ID Notes
OpenAI gpt-5.2 gpt-4o (until real API) GPT-5.2 flagship
Anthropic claude-opus-4.5 claude-opus-4-5 Most capable
Anthropic claude-sonnet-4.5 claude-sonnet-4-5 Balanced
Anthropic claude-haiku-4.5 claude-haiku-4-5 Fast/cheap
Google gemini-3-pro-preview gemini-3-pro-preview Multimodal flagship
Google gemini-2.5-pro gemini-2.5-pro Best quality (2M ctx)
Google gemini-3.1-flash-lite-preview gemini-3.1-flash-lite-preview Fast (1M ctx)

Models to DELETE (16 Total)

OpenAI (8 to delete)

  • gpt-5.1 - Not GPT-5.2 series
  • gpt-5.1-codex - Not GPT-5.2 series
  • gpt-5-mini - Not GPT-5.2 series
  • gpt-5-nano - Not GPT-5.2 series
  • gpt-4.1 - Old 4.x series
  • gpt-4.1-mini - Old 4.x series
  • gpt-4.1-nano - Old 4.x series
  • gpt-4o - Old 4.x series (keep as SDK mapping target only)

Anthropic (5 to delete)

  • claude-sonnet-4-5-20250929 - Use claude-sonnet-4.5 (no date suffix)
  • claude-opus-4-5-20251101 - Use claude-opus-4.5 (no date suffix)
  • claude-haiku-4-5-20251001 - Use claude-haiku-4.5 (no date suffix)
  • claude-sonnet-4 - Old Claude 4.0 series
  • claude-opus-4 - Old Claude 4.0 series

Read the full file on GitHub · 1,419 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 · 1,419 lines · 0 tokens per session scan A 88e9759c32bd

Subscribe to this mod's changes

MODEL_CONSOLIDATION_PLAN is an agent published in the GitHub repository HomenShum/NodeBenchAI (14 stars, last pushed 19d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 13,587 tokens. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.