# Deep Research Librarian > Deep Research Librarian is an AI agent skill for Claude Code, Codex CLI, Gemini CLI, and other agents that read SKILL.md. Finds the playbooks, courses, videos, and practitioners on a topic and files them. Best for: deep research skill for Claude. - URL: https://atharvashah.com/ai-skills/deep-research - Price: $19, one payment - Works with: Claude Code, Codex CLI, Gemini CLI, Cursor, GitHub Copilot - Category: Research and knowledge - Built by: Atharva Shah (https://atharvashah.com) - Requirements: Needs a paid Claude plan with web search turned on. Claude Code saves the file straight to your folder. Paywalled sources are linked and marked, and you decide whether to pay for them. - Buy: https://buy.polar.sh/polar_cl_UrbGOx20Q8jkmsQZp6DiawITRxXonNK11YhGk4UcDzG?custom_field_data.skill-standard=deep-research (the skill folder and setup guide arrive by email) ## A practitioner-grade reference file on any topic you name. It runs 15 or more searches, drops the shallow listicles, and files 25 to 30 sourced tactics, the people worth following, and a 30-day plan. ## What it produces - **The tactics file.** You get one reference file of sourced tactics, each with what to do and why it works. ![A real research file on agent orchestration, tilted, showing three numbered tactics with what to do and a named source.](https://atharvashah.com/ai-skills/deep-research/scene-1.webp) - **The 30-day plan.** Each file ends with a week-by-week plan you can start on Monday. ![The 30-day plan section of a real research file, with four weeks of concrete tasks.](https://atharvashah.com/ai-skills/deep-research/scene-2.webp) - **People to follow.** You learn which practitioners to follow and what each one is best at. ![The People to Follow section of a real research file, with five named practitioners and their handles.](https://atharvashah.com/ai-skills/deep-research/scene-3.webp) ## Name the topic. Get the file you would have built in a week. It asks one question if the angle is unclear, then does the digging. 1. **Name what you want to master.** One line is enough. If the angle is unclear, it asks whether you want tactics, theory, tooling, or all three. 2. **It searches and filters.** It covers playbooks, courses, videos, community threads, and downloads. Each source gets checked for date, depth, a named author, and real evidence. 3. **Open the reference file.** One markdown file lands in your resources folder, with a mental model, sourced tactics, a learning path, people to follow, and a week-by-week plan. ## One-click research reports read well and teach little. They summarize the first page of results. This skill filters for practitioners and files tactics you can run next week. ### Deep Research Librarian - **Tactics first.** Each tactic has a context, a timeline, a mechanism, and a source. - **Takes a side.** When sources conflict, it says which one it trusts and why. - **Real links or a flag.** A missing link says so, and paywalls are marked. - **A file you keep.** Every topic becomes one file in your own library. ### One-click AI reports - **Summary prose.** Long paragraphs that restate what the top results said. - **Neutral on purpose.** Two sources disagree and you get both, with no call. - **Loose citations.** A link that does not say what the sentence claims. - **Gone next week.** The report lives in a chat you will not find again. ### Doing it yourself - **Forty open tabs.** Half are listicles that repeat each other. - **No filter.** A 2019 post and a 2026 teardown look the same in results. - **Notes nowhere.** The good finds end up in bookmarks you never reopen. - **A lost weekend.** The research eats the time you meant for practice. ## Frequently asked questions ### Does Claude have a deep research mode? Claude has a research feature in its apps. This skill brings a repeatable method to Claude Code and Codex: a minimum search count, a filter for shallow sources, and a saved reference file you keep. ### What does the file contain? A mental model, the fundamentals, 25 to 30 sourced tactics, an AI and automation section, a learning path, downloads, people to follow, a 30-day plan, open questions, and a full source list. ### Does it make up sources? No. If it cannot find a direct link, it says so and gives you the exact title to search for. ### How long does one run take? Usually several minutes, since it runs at least 15 searches and reads what it keeps. Broad topics take longer. ### Which Claude plan do I need? A paid plan with web search. Claude Code writes the file to your folder directly. Pro handles a few topics a week. ### What else do I pay for? Nothing is required. Some courses and reports it finds are paid, and it marks those clearly. ### Does it work for niche topics? Yes, though a niche topic has fewer strong sources. It keeps five good ones over thirty weak ones and tells you where coverage was thin. ### Is it neutral? No, by design. When sources conflict, it says which one it trusts more and why. ### Where does the file go? Into a resources folder you choose at setup, named after the topic, so your library stays in one place. ### What if the first file misses my angle? Say which angle matters, such as tooling only or enterprise only, and run it again. The one clarifying question exists to catch this early. ### Does this skill work with Gemini CLI, Codex, Cursor, or GitHub Copilot? Yes, with one caveat. The skill is a SKILL.md folder in the Agent Skills open standard, which Claude Code, Codex CLI, Gemini CLI, Cursor, and GitHub Copilot all load. It is built and tested on Claude Code. Steps that lean on Claude-only features, such as parallel subagents, run one after another or need a manual step in other agents. ### How do I install an agent skill? Copy the skill folder into your agent's skills directory: ~/.claude/skills for Claude Code, or the skills folder your agent documents for Codex CLI, Gemini CLI, Cursor, or Copilot. The agent reads the description and loads the skill when your request matches it. Each skill ships with a setup guide for any tool it needs. [All AI skills](https://atharvashah.com/skill-pack) ยท [Library index](https://atharvashah.com/ai-skills.md)