The "Intelligent Package Manager" (IPM) is workflow tool to help manage your agentic artifacts. Combining industry standard "external" artifacts and internal ones private to you, it allows you to search, maintain and distribute "plugins" to target deployment shapes.
If you are like me I find navigating the latest and greatest tiresome, especially when I am trying to build packs for specific repos and projects. IPM allows you to choose the repos you are interested in, build a search database from them, then independently build packs (think Claude Code Plugins) you can install where ever you want. We provide a list of suggested repos, sourced from agentic-tools, you are free to use or find your own.
With a curated list of repos you can build your search database to allow you to easily find and navigate the artifacts, then build relevant packs for you and your team. Packs can live globally in user scope or locally in the project you are in.
Search across the catalog:
Browse the tree:
They can then be installed to target destinations: Claude Code, Gas City, Codex (coming soon), Cursor (coming soon).
Curated catalog of 344 GitHub repos for building software with AI agents, in 24 lists: coding agents, skills, MCP, orchestration, memory. Searchable, filterable GitHub Pages site plus CSV, TOML and JSON exports. Stdlib Python build, vanilla JS, no dependencies.

AI-assisted form builder: upload a PDF or describe a form, get a structured, publishable web form with computed values & conditional logic (jexl). TanStack Start + React 19 on Cloudflare Workers, Neon Postgres, Better Auth, Python/LangGraph extraction.
ezc (easy config): one git-backed store for secrets and variables, layered by environment into every repo. Encrypts secrets with SOPS + age, manages who can read them (join requests, user and CI bot keys), validates each repo's environment against a manifest and loads it through mise, direnv or ezc run. A Go CLI built with Cobra.
Recipes, a reusable TOML config pack, and copy-me example cities for Gas City multi-agent coding workflows. Agents ask for a model tier (fast, solid, deep); each city binds it to Claude, Gemini (Antigravity) or Kimi.
21 ready-made stacks for
Daily tech-news digest built by a LangGraph swarm of 5 AI agents: collects from 20+ sources (RSS, Reddit, Dev.to, GitHub Trending, Lobsters), LLM-scores on 6 dimensions, summarises via Chain of Density, and publishes a JSON newsfeed + images to Cloudflare R2. Also runs the Obsidian-to-R2 blog ETL. Python, Click, Pydantic, mise/uv, GitHub Actions.