By name: quick-research description: Single-pass web research for fast factual lookups in about 2 minutes. USE WHEN quick research, quick lookup, minor research, fast lookup, simple factual question, quick check.
Quick Research
Mode: Single-pass web research | Time: ~2 minutes
When to Use
- User says "quick research", "quick lookup", or "minor research"
- Simple factual questions needing web verification
- Time-sensitive requests where speed matters more than depth
- Surface-level understanding of a topic before deciding on deeper research
Available Tools
| Tool | Use For |
|---|---|
| WebSearch | Fast web queries, returns search result snippets with URLs |
| WebFetch | Deep read of a specific URL, extracts and processes page content |
Workflow
Step 1: Formulate Query
Analyze the user's request and craft ONE focused search query.
Good queries are:
- Specific and well-scoped (not overly broad)
- Include key terms that will surface authoritative results
- Framed to get factual answers, not opinion
Example: User asks "What's the current state of WebGPU adoption?"
- Query: "WebGPU browser adoption status 2026"
Step 2: Execute Search
Run a single WebSearch call with the crafted query.
WebSearch(query: "[your focused query]")
Review the returned snippets. Identify the 1-2 most relevant and authoritative results.
Step 3: Deep Read (If Needed)
If the search snippets alone do not provide a confident answer, use WebFetch on the most promising result URL to get full content.
WebFetch(url: "[best result URL]", prompt: "Extract the key facts about [topic]")
Skip this step if the search snippets already contain a clear, confident answer.
Step 4: Assess Confidence
Rate your confidence in the findings:
- HIGH -- Multiple consistent sources, recent data, authoritative publishers
- MEDIUM -- Single source or partially dated information
- LOW -- Conflicting sources, no authoritative backing, or sparse results
Step 5: Deliver Results
Present findings in this format:
## Quick Research: [Topic]
**Query:** [what was searched]
**Confidence:** [HIGH / MEDIUM / LOW]
### Key Findings
- [Finding 1 with source attribution]
- [Finding 2 with source attribution]
- [Finding 3 if applicable]
### Sources
- [Source title](URL) -- [one-line relevance note]
### Recommendation
[If confidence is LOW or MEDIUM, suggest standard-research for deeper coverage]
Constraints
- Do NOT run more than 2 WebSearch queries. This is quick mode.
- Do NOT use WebFetch on more than 2 URLs. Pick the best sources.
- If the topic requires multi-perspective analysis, recommend standard-research instead.
- Total elapsed time target: under 2 minutes.
Output Persistence
Quick research results are delivered inline to the user. They are NOT automatically persisted to disk. If the user asks to save the results, write them to:
~/.augment/MEMORY/RESEARCH/{YYYY-MM-DD}_{topic-slug}/quick-research.md
Escalation
If quick research reveals the topic is more complex than expected, tell the user:
"This topic has more depth than a quick lookup can cover. I recommend running standard research for a comprehensive view."
Do not automatically escalate -- let the user decide.