1---
2name: tavily-research
3description: |
4  Conduct comprehensive AI-powered research with citations via the Tavily CLI. Use this skill when the user wants deep research, a detailed report, a comparison, market analysis, literature review, or says "research", "investigate", "analyze in depth", "compare X vs Y", "what does the market look like for", or needs multi-source synthesis with explicit citations. Returns a structured report grounded in web sources. Takes 30-120 seconds. For quick fact-finding, use tavily-search instead.
5allowed-tools: Bash(tvly *)
6---
7 
8# tavily research
9 
10AI-powered deep research that gathers sources, analyzes them, and produces a cited report. Takes 30-120 seconds.
11 
12## Before running any command
13 
14If `tvly` is not found on PATH, install it first:
15 
16```bash
17curl -fsSL https://cli.tavily.com/install.sh | bash && tvly login
18```
19 
20Do not skip this step or fall back to other tools.
21 
22See [tavily-cli](../tavily-cli/SKILL.md) for alternative install methods and auth options.
23 
24## When to use
25 
26- You need comprehensive, multi-source analysis
27- The user wants a comparison, market report, or literature review
28- Quick searches aren't enough — you need synthesis with citations
29- Step 5 in the [workflow](../tavily-cli/SKILL.md): search → extract → map → crawl → **research**
30 
31## Quick start
32 
33```bash
34# Basic research (waits for completion)
35tvly research "competitive landscape of AI code assistants"
36 
37# Pro model for comprehensive analysis
38tvly research "electric vehicle market analysis" --model pro
39 
40# Stream results in real-time
41tvly research "AI agent frameworks comparison" --stream
42 
43# Save report to file
44tvly research "fintech trends 2025" --model pro -o fintech-report.md
45 
46# JSON output for agents
47tvly research "quantum computing breakthroughs" --json
48```
49 
50## Options
51 
52| Option | Description |
53|--------|-------------|
54| `--model` | `mini`, `pro`, or `auto` (default) |
55| `--stream` | Stream results in real-time |
56| `--no-wait` | Return request_id immediately (async) |
57| `--output-schema` | Path to JSON schema for structured output |
58| `--citation-format` | `numbered`, `mla`, `apa`, `chicago` |
59| `--poll-interval` | Seconds between checks (default: 10) |
60| `--timeout` | Max wait seconds (default: 600) |
61| `-o, --output` | Save output to file |
62| `--json` | Structured JSON output |
63 
64## Model selection
65 
66| Model | Use for | Speed |
67|-------|---------|-------|
68| `mini` | Single-topic, targeted research | ~30s |
69| `pro` | Comprehensive multi-angle analysis | ~60-120s |
70| `auto` | API chooses based on complexity | Varies |
71 
72**Rule of thumb:** "What does X do?" → mini. "X vs Y vs Z" or "best way to..." → pro.
73 
74## Async workflow
75 
76For long-running research, you can start and poll separately:
77 
78```bash
79# Start without waiting
80tvly research "topic" --no-wait --json    # returns request_id
81 
82# Check status
83tvly research status <request_id> --json
84 
85# Wait for completion
86tvly research poll <request_id> --json -o result.json
87```
88 
89## Tips
90 
91- **Research takes 30-120 seconds** — use `--stream` to see progress in real-time.
92- **Use `--model pro`** for complex comparisons or multi-faceted topics.
93- **Use `--output-schema`** to get structured JSON output matching a custom schema.
94- **For quick facts**, use `tvly search` instead — research is for deep synthesis.
95- Read from stdin: `echo "query" | tvly research - --json`
96 
97## See also
98 
99- [tavily-search](../tavily-search/SKILL.md) — quick web search for simple lookups
100- [tavily-crawl](../tavily-crawl/SKILL.md) — bulk extract from a site for your own analysis
101