research-tasks

Answers research questions using web search and saves findings as markdown notes.

27.6k|2.7k|Updated Aug 6, 2024
One-click install
npx skills add https://github.com/mastra-ai/mastra --skill research-tasks
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: research-tasks
Source: https://github.com/mastra-ai/mastra/tree/main/templates/template-claw-assistant/workspace/skills/research-tasks
Command: npx skills add https://github.com/mastra-ai/mastra --skill research-tasks

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Answering research questions often leads to unsourced claims and lost references. This Skill enforces a disciplined process for web research that captures sources, quotes, and notes so every answer is verifiable and reproducible.

Core Features & Use Cases

  • Structured Web Research: Uses web_search for broad queries and browser tools for reading specific pages in depth.
  • Persistent Notes: Saves useful URLs and short quotes into markdown files under a notes/ directory in the workspace.
  • Source Integrity: Forbids invented sources; any claim not backed by a fetched page is dropped from the answer.
  • Use Case: Ask a question like "What are the current pricing tiers of vector database providers?" and receive a one-paragraph answer, a bulleted list of source URLs, and a saved markdown notes file for later reference.

Quick Start

Research the latest developments in small language models and save your findings with sources to my workspace notes.

Frequently Asked Questions about research-tasks

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I answer research questions with cited web sources?

Start with web_search for broad queries, then use browser tools to read specific pages. Capture URLs and short quotes into a markdown notes file, and deliver a one-paragraph answer with a bulleted source list.

When should I use web search versus browser tools for research?

Use web_search first for broad queries to discover relevant pages. Switch to browser tools when you need to read the full content of a specific page found during the search phase.

How do I prevent AI from inventing sources in research answers?

Enforce a rule that any claim not supported by an actually fetched page must be dropped. This Skill explicitly forbids invented sources and only cites URLs captured during real browsing.

Where are research findings saved in the workspace?

Findings are saved as markdown files under the notes/ directory in the workspace. Each notes file contains captured URLs and short quotes gathered during the research session.

What format does the final research answer use?

The final answer has three parts: a one-paragraph Answer section, a bulleted Sources list of URLs, and a Notes file path pointing to the saved markdown notes.