iPSS Agent

InterPSS Agentic Power System Simulation Agent for AC load flow, DC-based contingency analysis, and NERC TPL-001-5 style reporting. This repository ships agent-facing skills for OpenAI Codex Desktop, Claude Code CLI, and DeepSeek Harness (build and CLI details in Setup.md; DSH plugin in InstallDSHPlugin.md). The canonical skill is [.agents/skills/ipss-sim/SKILL.md](.agents/skills/ipss-sim/SKILL.md); run ./scripts/sync_ipss_skills.sh after edits to refresh .claude/skills/ipss-sim/SKILL.md.

It is also integrated into DeepSeek Harness as a DSH Plugin. You can run power system simulation in the traditional step-by-step way (plugin user guide):

Or in the native AI Chat way (agent chat user guide):

Environment setup

Prerequisites: Java JDK 21, Maven (or the included mvnw wrapper).

Prompt Codex, Claude Code, or DeepSeek Harness to clone and build ipss-agent:

Install(Update) and setup ipss-agent from https://github.com/InterPSS-Project/ipss-agent

Depending on your network speed, it may take some time to download dependencies and build the CLI.

Build the Java CLI from the project root:

./mvnw -q clean package

This produces target/ipss-agent-cmd-1.0.0-uber.jar. See Setup.md for the full layout, tests, and configuration.

Test the setup with a sample case directory:

init  # for the first time installation
/ipss-sim data/ieee/Ieee118Bus/ "IEEE 118-Bus Test Case"

DSH Plugin setup

Prompt DeepSeek Harness to install InterPSS DSH plugin:

Install(update) InterPSS DSH plugin

User Guide

• InterPSS DSH Plugin user guide plugin_user_guide.md, plugin_user_guide.pdf
• iPSS Agent Chat user guide agent_chat_user_guide.md

Reference

Topic	Document
Layout, build, JARs, agent skills	Setup.md
IpssCmd (Java CLI) usage	IpssCmd.md
Markdown report generator	GenReport.md
DeepSeek Harness DSH plugin	InstallDSHPlugin.md
Interactive HTML dashboards	[.agents/skills/nerc-report-html/SKILL.md](.agents/skills/nerc-report-html/SKILL.md)
NERC slide decks	[.agents/skills/nerc-report-slides/SKILL.md](.agents/skills/nerc-report-slides/SKILL.md)