by iusztinpaul
How to turn your Second Brain into a living research memory that your agents maintain. Workshop with slides, video and code.
# Add to your Claude Code skills
git clone https://github.com/iusztinpaul/ai-research-os-workshopGuides for using ai agents skills like ai-research-os-workshop.
ai-research-os-workshop is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by iusztinpaul. How to turn your Second Brain into a living research memory that your agents maintain. Workshop with slides, video and code. It has 51 GitHub stars.
ai-research-os-workshop's catalog security scan is still queued. You can run an instant dependency and prompt-injection check now with the "Scan for vulnerabilities" button above.
Clone the repository with "git clone https://github.com/iusztinpaul/ai-research-os-workshop" and add it to your Claude Code skills directory (see the Installation section above).
ai-research-os-workshop is primarily written in Python. It is open-source under iusztinpaul on GitHub, so you can review or fork the full source.
Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh ai-research-os-workshop against similar tools.
No comments yet. Be the first to share your thoughts!
Unlocks once the catalog security scan passes (runs nightly).
The deep catalog scan for this skill is still queued. Run an instant dependency check now instead.
I'm always losing my research.
5,000+ notes in Obsidian, 5,000+ highlights in Readwise, plus Notion and Google Drive — growing ~250 files a month. And every research session still starts from zero: paste the same links into Codex, watch it rebuild context on the fly, then lose all of it when the chat ends.
Access to information was never the bottleneck. Reusing it was.
AI Research OS turns your Second Brain into a living research memory your agents maintain —
research that compounds over weeks, months, and years instead of dying in a conversation.
Codex or Claude gives you an answer. This gives your harness a reusable research memory:
raw/ - copied or fetched source materialwiki/sources/ - per-source summarieswiki/concepts/, wiki/entities/, wiki/comparisons/ - reusable synthesis pageswiki/overview.md and wiki/synthesis.md - the current thesisindex.yaml and index.md - the catalog future agents read firstlog.md and wiki/open-questions.md - what happened and what to research nextThe point is not to replace Codex. The point is to stop re-researching the same topic every week.
AI Research OS is a set of local AI skills for building and querying a persistent research wiki from your own sources:
Via deep research on top of your personal Second Brain
Obsidian is optional. It is just a visual IDE for browsing the generated markdown wiki. The system can run purely through Codex or Claude Code from a normal working directory.
Our AI Research OS runs locally via skills.
Our Agent AI Engineering course, built with Towards AI, shows how to ship it to production as a multi-agent system: MCP servers with LangGraph, an evaluator-optimizer loop, observability, evals, and GCP deployment.
35 lessons. 3 end-to-end portfolio projects. A certificate. And a Discord community with direct access to industry experts and me.
Built for software engineers, data engineers, or scientists transitioning into AI engineering.
Rated 5/5 by 300+ students. The first 7 lessons are free:
Sources flow through deep research, get stored as raw files, indexed, synthesized into a wiki, and then queried:
user question / sources
|
v
/research router
|
+--> query existing wiki
+--> append known sources
+--> run deep discovery
|
v
raw sources -> source pages -> concepts/entities/comparisons
|
v
index.yaml + overview.md + synthesis.md + open-questions.md
Three end-to-end runs, each browsable in examples/. Open the linked prompt
in Claude Code / Codex with /research to reproduce it; each screenshot shows the resulting
wiki browsed in Obsidian.
Discover sources, summarize, and synthesize them into a topic wiki.
How to use it: /research example_1_deep_research/prompt.md
A full research wiki on agentic harnesses, built from a content outline and reference links.
Compute per-repo notes (architecture, agents, memory, permissions) and cross-repo comparisons, skipping deep discovery.
How to use it: /research example_2_github/prompt.md
Side-by-side comparison pages across three coding-agent repositories.
Pull a handful of specific articles into the wiki without running deep research.
How to use it: /research example_3_ingest_links/prompt.md
Source pages and synthesis built from three custom URLs.
/research.working-dir/research-<topic>/ directory.Use this when:
Do not use this when:
| Tool | Best for | Limitation | Where AI Research OS fits |
|---|---|---|---|
| Codex one-shot | Fast answers, coding help, repo Q&A | The answer is not automatically turned into a durable research workspace | Use Codex directly for simple questions; use this when the research should be reused and extended |
| NotebookLM | Chatting with a fixed set of uploaded sources | Less programmable, less agent-native, not designed around repo parsing, wiki updates, or repeated source ingestion loops | Creates local files, source pages, indexes, and synthesis that agents can keep editing |
| Deep research agents | Broad discovery and synthesis | Often produce a one-time report | Stores the report as a living wiki with raw sources, open questions, and append workflows |
| RAG / vector databases | Retrieval over large corpora | Infrastructure-heavy; retrieval alone does not create source pages, comparisons, or a thesis | Keeps the workflow lightweight and artifact-first; indexing is human/agent-readable |
AI Research OS |
Research that compounds across notes, repos, videos, links, and follow-up questions | More setup than a one-shot prompt | Gives Codex / Claude Code a reusable research workspace |
| Skill | What it does |
|---|---|
/research |
Init, append, or query a per-topic research directory. |
/research-distill |
Distill a research directory into a compact research.md for a specific piece of content. |
/research-lint |
Health-check a research directory for orphan sources, broken links, stale claims, contradictions, and missing hubs. |
/research-render |
Render wiki pages into slides, charts, canvases, or content briefs. |
The shared data contract lives in
plugins/ai-research-os/skills/research/CONVENTIONS.md.
/research routes requests before doing expensive work:
| Mode | Use when | Behavior |
|---|---|---|
query |
Ask from an existing research directory | Reads index.yaml and wiki/; no ingest or discovery. |
append |
Add the sources you provide | Ingests the provided sources only; no discovery rounds. |
deep |
Explicitly request deep research | Runs source discovery rounds (at a fast/light/deep depth preset), dedup, and wiki updates. |
init |
Start a new research directory | Creates working-dir/research-<topic>/. |
Deep discovery is opt-in and runs at one of three depth presets — fast (1 round, 3 queries),
light (2 rounds, 3 + 2 queries), or deep (3 rounds, 3 queries each). Long runs show a plan
first: selected mode, sources to ingest, expected runtime, and files to write.
In query mode the agent drills down progressively — index summary first, then the source
wiki page, then derivatives, and only the raw source if it still needs it — so the context
window stays small: