AI agent infrastructure

Practical infrastructure for AI coding agents.

I build open-source tools for controlling, inspecting, continuing, and extending AI coding-agent workflows across Pi, Cursor, Codex-style sessions, browsers, and web-model handoffs.

Current public suite: 1,058 GitHub stars across 5 agent-workflow tools, with 40,675 npm package downloads last month.

Explore tools View GitHub Contact

Open-source suite5focused agent-workflow tools
GitHub signal1,058stars across the tracked suite
npm package signal40,675downloads last month across the tracked suite
npm dataOctober 4, 2026 MDTlast verified from npm package APIs

Lab console

GitHub metrics refresh from public APIs; npm downloads last verified October 4, 2026 MDT.

Repository & npm package activity

pi-posthorse
stars 255forks 7npm 0.8.0npm 1,344/mo
pi-cursor-sdk
stars 339forks 75npm 0.4.0npm 9,482/mo
pi-agent-browser-native
stars 230forks 38npm 0.9.2npm 24,426/mo
pi-codex-goal
stars 195forks 29npm 0.6.0npm 4,816/mo
pi-oracle
stars 39forks 11npm 0.7.21npm 607/mo

Current project notes

01Fresh context without summariesContext rollover with durable notes and searchable history.
02Cursor models inside PiProvider integration and Pi-native tool activity replay.
03Browser automation as an agent toolSnapshots, clicks, screenshots, downloads, QA, and Electron support.
04Codex-style goals for long-running sessionsGoal state and continuation context across long agent work.
05Async second-model review from PiContext-rich ChatGPT/Grok handoffs without blocking the session.

source: GitHub, npm, and project repositories.

Featured agent workflow tools

Public proof for the infrastructure layer around coding agents: context rollover, provider integration, browser automation, goal state, and model handoffs across Pi, Cursor, Codex-style sessions, and web workflows.

Browse all tools →

pi-posthorse

GitHub ↗npm ↗

No-summary context rollover for Pi, with durable notes and history recovery.

Stars: 255
Forks: 7
npm downloads: 1,344/mo

Workflow gap: Long agent sessions hit the context limit, and summary-based compaction replaces the work so far with a lossy recap.
What it adds: Moves the agent to a fresh context window with a handoff it writes itself, calls no summary model, keeps the full transcript searchable, and adds notes that survive resets.

Pi
context windows
session history
durable notes

pi / cursor

pi-cursor-sdk

GitHub ↗npm ↗

Cursor SDK provider extension for Pi.

Stars: 339
Forks: 75
npm downloads: 9,482/mo

Workflow gap: Cursor-backed agents and models should be usable from the same controlled workflow surface as other coding-agent providers.
What it adds: Brings Cursor models into Pi’s provider/model flow, with model-picker integration, local Pi tool bridging through MCP, and Pi-native tool activity replay.

Pi
Cursor
provider integration
MCP
model workflow

pi / browser

pi-agent-browser-native

GitHub ↗npm ↗

Native browser and desktop-app automation for Pi agents.

Stars: 230
Forks: 38
npm downloads: 24,426/mo

Workflow gap: Real agent work often needs browser context, screenshots, clicks, downloads, QA checks, and app inspection without leaving the coding session.
What it adds: Exposes agent-browser as a native Pi tool with browser automation, compact snapshots, screenshots, downloads, QA presets, source lookup, and Electron lifecycle support.

Pi
browser automation
Electron
QA
tooling

pi / codex

pi-codex-goal

GitHub ↗npm ↗

Codex-style goal tracking and continuation for long-running Pi sessions.

Stars: 195
Forks: 29
npm downloads: 4,816/mo

Workflow gap: Long-running coding-agent sessions lose their objective, drift across compaction, and become hard to resume cleanly.
What it adds: Adds a concrete goal object, token/progress visibility, and continuation context so agent work can stay anchored across long sessions.

Pi
Codex
goal tracking
continuation
session state

pi-oracle

GitHub ↗npm ↗

ChatGPT/Grok web-oracle extension for Pi with isolated browser auth and async jobs.

Stars: 39
Forks: 11
npm downloads: 607/mo

Workflow gap: Hard coding decisions sometimes need another model with full project context without blocking the main agent session.
What it adds: Packages project context, submits web-model jobs, tracks async status, archives artifacts, and reads results back into the workflow.

Pi
ChatGPT
Grok
async jobs
model handoff

See the full tool inventory →

Older field notes

Useful notes and essays live here, but current agent-tooling proof is above in the tools and public activity sections.

Browse field notes →

Treat Your Agent Logs As Training Data

How I run performance reviews on Codex and Claude using their own session history.

November 27, 2025

AI Agents Codex

Effective Prompting Strategies: Quick Reference with Reusable Examples

A quick reference guide that provides reusable examples and best practices for prompting advanced LLMs in coding, troubleshooting, creative writing, and more.

March 6, 2025

AI Guide

Protesting AI Detection Rulings: A Mini-Guide

A practical guide for students and professionals who have been wrongly accused of using AI to generate their original work. This article outlines evidence-based strategies to contest false AI detection flags, including documenting your process, comparing writing samples, citing detection limitations, and proposing reasonable resolutions.

March 3, 2025

AI Education Guide

Security and data notes

A supporting research surface for practical security data and exploited-vulnerability inspection.

Research surface

CISA Known Exploited Vulnerabilities

Search and inspect the KEV catalog by CVE, vendor, product, timeline, and ransomware use.

Open KEV explorer →