deep-research
Conduct comprehensive, multi-source research and generate cited reports. Activate this skill when users want in-depth research on a topic, need synthesis across web and vault sources, or want a structured research report saved to their vault.
Best use case
deep-research is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
Conduct comprehensive, multi-source research and generate cited reports. Activate this skill when users want in-depth research on a topic, need synthesis across web and vault sources, or want a structured research report saved to their vault.
Teams using deep-research should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/deep-research/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How deep-research Compares
| Feature / Agent | deep-research | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
Conduct comprehensive, multi-source research and generate cited reports. Activate this skill when users want in-depth research on a topic, need synthesis across web and vault sources, or want a structured research report saved to their vault.
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
SKILL.md Source
# Deep Research Conduct comprehensive, multi-round research using the `deep_research` tool. This produces structured markdown reports with citations by iteratively searching, reading, and synthesizing multiple sources. ## When to Use Deep Research vs Google Search | Use `deep_research` when... | Use `google_search` when... | | ---------------------------------------------------------- | ------------------------------------------- | | The topic requires synthesis across multiple sources | A quick factual answer is needed | | The user wants a structured report or overview | The user asks for a single data point | | Research should combine vault notes with web sources | The user wants current news or events | | The question is broad or exploratory | The question is specific and narrow | | The user explicitly asks for "deep research" or "a report" | The user asks "what is..." or "when did..." | **Default to `google_search` for simple questions.** Only use `deep_research` when the scope genuinely warrants multi-source synthesis. ## How to Use Call the `deep_research` tool with these parameters: - **`topic`** (required) — The research question or topic. Be specific and descriptive. - **`scope`** (optional) — Where to search: - `both` (default) — Web sources + vault notes. Best for most research. - `web_only` — Internet sources only. Use when the topic is outside the vault's scope. - `vault_only` — Vault notes only. Use to synthesize the user's own notes. Requires Semantic Vault Search to be enabled. - **`outputFile`** (optional) — Path to save the report (e.g., `Research/topic-name.md`). A `.md` extension is added automatically if missing. If `both` is requested but Semantic Vault Search isn't configured, the tool gracefully falls back to web-only research. ## Best Practices - **Always suggest an `outputFile`** — Research reports are valuable artifacts. Suggest saving to a logical location like `Research/` or a project folder. - **Warn the user about timing** — Deep research takes several minutes. Let the user know before invoking the tool. - **Choose the right scope**: - User wants to understand a new external topic → `web_only` or `both` - User wants to synthesize their own notes → `vault_only` - User wants to combine personal knowledge with external research → `both` - **Write a clear topic** — Frame the topic as a research question or focused description. "The evolution of transformer architectures in NLP" is better than "transformers". ## Output Format The tool generates a markdown report with: - A title and generation date - Synthesized findings organized into logical sections - Inline citations linking to source URLs - Source count summary When `outputFile` is specified, the report is saved to the vault and added to the session context for follow-up discussion. ## Tips - After receiving a research report, offer to help the user extract action items, create summary notes, or link findings to existing vault content. - For iterative research, suggest running multiple focused queries rather than one broad one. - If the user wants to research something related to their existing notes, suggest `both` scope to combine external knowledge with their personal context.
Related Skills
vault-semantic-search
Search vault notes by meaning using semantic search (RAG). Activate this skill when users want to find notes by concept or topic rather than exact keywords, or when keyword search tools return poor results.
recall-sessions
Search past agent conversations to recall prior discussions, decisions, and context. Activate this skill when users ask about previous conversations, want to resume past work, or reference earlier decisions.
obsidian-properties
Work with Obsidian note properties (frontmatter). Activate this skill when users want to add, modify, or organize properties, understand property types, format YAML frontmatter, or use properties with templates, search, or Bases.
obsidian-bases
Create and configure Obsidian Bases — database-like views of notes. Activate this skill when users want to create bases, write filters, formulas, or set up table/cards/list/map views.
image-generation
Generate images from text descriptions and save them to the vault. Activate this skill when users want to create illustrations, diagrams, visual content, or any AI-generated images.
gemini-scribe-help
Answer questions about Gemini Scribe plugin features, settings, and usage. Activate this skill when users ask how to use the plugin, configure settings, or troubleshoot issues.
audio-transcription
Transcribe audio and video files into structured notes. Activate this skill when users want to transcribe recordings, meetings, podcasts, voice memos, or any audio/video content in their vault.
ui-ux-guidelines
UI/UX best practices for obsidian-gemini plugin development. Covers modal sizing, text overflow, message formatting, collapsible UI, animations, icons, file chips, session state, CSS containment, and theme compatibility. Use this skill when building or modifying UI components.
release-process
Full release workflow for obsidian-gemini: update release notes, run checks, bump version with npm, create a GitHub release, and verify. Use this skill when preparing a new plugin release.
obsidian-plugin-development
Build, modify, and debug Obsidian plugins using the TypeScript API. Use this skill when working with Obsidian plugin source code, the obsidian npm package, plugin UI (views, modals, settings, commands, ribbons), vault file operations, editor manipulation, workspace management, metadata cache, events, markdown rendering, or the Obsidian CLI. Covers plugin lifecycle, best practices, common patterns, and the full TypeScript API surface.
obsidian-cli
Use the Obsidian CLI to debug, inspect, and test Obsidian plugins during development. Covers plugin reloading, console inspection, runtime evaluation, and common debugging recipes for the gemini-scribe plugin.
gemini-api-dev
Use this skill when building applications with Gemini models, Gemini API, working with multimodal content (text, images, audio, video), implementing function calling, using structured outputs, or needing current model specifications. Covers SDK usage (google-genai for Python, @google/genai for JavaScript/TypeScript, com.google.genai:google-genai for Java, google.golang.org/genai for Go), model selection, and API capabilities.