# Benchmark CLI agents on autonomous LLM post-training with PostTrainBench

> Run Claude Code, Codex CLI, Gemini CLI, or OpenCode through bounded H100 post-training tasks and compare how well each agent improves a base LLM.

- Skill: `agentskillexchange/benchmark-cli-agents-on-autonomous-llm-post-training-with-po` (Agent Skill)
- Install (CLI): `npx skillmds@latest add agentskillexchange/benchmark-cli-agents-on-autonomous-llm-post-training-with-po`
- Raw SKILL.md: https://api.skillmd.com/api/skills/agentskillexchange/benchmark-cli-agents-on-autonomous-llm-post-training-with-po/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: agentskillexchange (https://skillmd.com/u/agentskillexchange)
- Updated: 2026-09-08
- Page: https://skillmd.com/skills/agentskillexchange/benchmark-cli-agents-on-autonomous-llm-post-training-with-po

---


# Benchmark CLI agents on autonomous LLM post-training with PostTrainBench

Run Claude Code, Codex CLI, Gemini CLI, or OpenCode through bounded H100 post-training tasks and compare how well each agent improves a base LLM.

## Prerequisites

Python, apptainer, fuse-overlayfs, Hugging Face cache, H100 GPU access, currently HTCondor scheduler support, and credentials for the selected CLI agent scaffolds

## Installation

Install or set up from the source-backed instructions:

Clone https://github.com/aisa-group/PostTrainBench, install requirements including apptainer and fuse-overlayfs, build the standard container with bash containers/build_container.sh standard, download the Hugging Face cache with bash containers/download_hf_cache/download_hf_cache.sh, copy example.env to .env, set API keys and paths, then submit jobs with bash src/commit_utils/commit.sh.

- Source: https://github.com/aisa-group/PostTrainBench

## Documentation

- http://posttrainbench.com/

## Source

- [Agent Skill Exchange](https://agentskillexchange.com/skills/benchmark-cli-agents-on-autonomous-llm-post-training-with-posttrainbench/)