plexe AGENTS.md

Repository instructions for Plexe, a framework that builds machine-learning models from plain-language requests. They describe its structure, workflow, commands, and cooperating agents.

In plain words
What is it for?
Use them when developing, running, or troubleshooting Plexe's data analysis, preparation, model-building, and Spark workflows.
Why use it?
They give coding agents the project-specific rules needed to understand and work on the repository correctly.

Instructions file for CodexOpenCode

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/plexe-ai/plexe/agents-md
Clone the repo
git clone --depth 1 https://github.com/plexe-ai/plexe

Made for: Codex, OpenCode.

Per session 1,431 This file is loaded in full into every session.
When invoked 1,431 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 $0.01431 $0.01431
Opus 5 $0.00715 $0.00715
Sonnet 5 $0.00286 $0.00286
Haiku 4.5 $0.00143 $0.00143

Measured yesterday against content hash fda2fe87dd86, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

plexe 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 yesterday.

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 · 123 lines

How it starts

The opening of the file, as written. The whole thing — 123 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AGENTS.md

This file provides guidance to coding agents when working with code in this repository (for example, OpenAI Codex and Claude Code).

AGENTS.md is the canonical version of this document. CLAUDE.md is kept as a compatibility symlink that points here, so if you opened this file via CLAUDE.md, you are in the right place.

Overview

Plexe is an agentic framework for building ML models from natural language. It employs a multi-agent architecture where specialized AI agents collaborate to analyze data, generate solutions, and build functional ML models through an autonomous 6-phase workflow.

Entry point: python -m plexe.main --train-dataset-uri <uri> --user-id <id> --intent "<task>" --spark-mode <local|databricks>

Docker:

  • docker build . — default PySpark image (local Spark execution)
  • docker build --target databricks . — Databricks Connect image (remote execution)

Architecture

6-Phase Workflow (plexe/workflow.py)

  1. Data Understanding: Statistical analysis → ML task identification → metric selection (with optional custom metric generation)
  2. Data Preparation: Dataset splitting → intelligent sampling (default: 30k train, 10k val samples)
  3. Baseline Models: Heuristic baseline with retry logic
  4. Model Search: Hypothesis-driven tree search on samples (iterative improvement)
  5. Final Evaluation: Optional test evaluation (default: disabled, uses validation performance)
  6. Packaging: Consolidates artifacts into work_dir/model/ (schemas/, config/, artifacts/, src/, evaluation/)

Multi-Agent System (plexe/agents/)

14 specialized agents orchestrate the workflow:

  • LayoutDetectionAgentStatisticalAnalyserAgentMLTaskAnalyserAgentMetricSelectorAgent (Phase 1)
  • MetricImplementationAgent: Generates custom metric code if not in StandardMetric enum
  • DatasetSplitterAgentSamplingAgent (Phase 2)
  • BaselineBuilderAgent (Phase 3)
  • HypothesiserAgentPlannerAgentFeatureProcessorAgent + ModelDefinerAgent (Phase 4 loop)
  • InsightExtractorAgent: Analyzes variant results, populates InsightStore for future hypotheses
  • ModelEvaluatorAgent: Multi-phase evaluation (Phase 5)

Read the full file on GitHub · 123 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. yesterday First seen · 123 lines · 1,431 tokens per session scan A fda2fe87dd86

Subscribe to this mod's changes

plexe AGENTS.md is an instructions file published in the GitHub repository plexe-ai/plexe (2,614 stars, last pushed 5mo ago), licensed Apache-2.0. It adds 1,431 tokens to every session, about $0.0072 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 2026-08-30.