Research

Skill SongYunSeop/oh-my-til/skills/research

Research a topic to identify key concepts and terms, then organize as a learning backlog. Use when the user wants a learning roadmap, study plan, curriculum, or asks 'what should I learn about X' — focuses on planning what to learn, not learning itself.From its SKILL.md

Install

npx -y skills add SongYunSeop/oh-my-til --skill research

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

3.3 KB, as published. Nobody here has run it

TIL_PATH: All ${TIL_PATH}/... references below resolve to the vault's configured TIL path (injected at SessionStart). Bash expands ${TIL_PATH} automatically. For Read/Edit/Glob/Write tools, substitute the resolved value before passing it as an argument (e.g. ${TIL_PATH}/SCHEMA.md → 40-Resources/til/SCHEMA.md). Default til when not injected; call MCP til_get_path if you need the current value.

Research Skill

Topic research → Identify concepts and dependencies → Save backlog file.

MCP Tools

til_list: Check existing TILs (search by topic using the search parameter)

Phase 1: Topic Research

Use til_list(search=topic) to check already-learned topics → avoid duplicate backlog entries
Research the topic via web search, identify required concepts, terms, and prerequisites
Break down into subtopics and research each directly
Analyze dependencies between subtopics

Phase 1.5: Save Raw Sources

Save high-quality sources to raw/ for future reference. Prioritize by source quality:

Save to raw/: official docs, original author posts (RFCs, maintainer blogs), in-depth technical articles with code examples
URL only (backlog sources, not saved to raw/): shallow overviews, listicles, undated posts, aggregator content
Skip: paywalled content, auto-generated pages, outdated material (2+ years for fast-moving topics)

For each source worth saving:

Save to raw/{category}/{slug}.md (auto-create folder)

Frontmatter format:

---
title: "Article Title"
source: https://original-url
fetched: YYYY-MM-DD
tags:
  - {category}
  - {subtopic}
---

Body: the full content retrieved via WebFetch, converted to markdown
Raw files are immutable — if raw/{category}/{slug}.md already exists, skip (do not overwrite)
Record saved raw file paths for use in Phase 2

Phase 2: Organize Backlog

Sort by learning order: Prerequisites → Core Concepts → Advanced
Add a one-line description per item
Gather user feedback: allow adding, removing, or reordering items

Phase 3: Save

Save to ./${TIL_PATH}/{category}/backlog.md (auto-create folder)
If backlog.md already exists, merge:
- Preserve [x] completed items
- Keep check state for matching items
- Preserve existing sources, only add new items
- sources values: use raw file paths (raw/{category}/{slug}.md) when saved in Phase 1.5, fall back to URLs for sources not saved locally
Add backlog link to TIL MOC
Commit all changes: 📋 research: {topic} learning backlog - {category} (do not push)

Backlog Template

Read references/templates.md for the exact backlog template and field descriptions.

Arguments

First: research topic (required)
Second: category (optional, automatically detected if omitted)

Rules

One-line description per item, split into multiple files if over 20 items (long backlogs hurt readability and cause scroll fatigue)
Links: [display name](${TIL_PATH}/{category}/{slug}.md)