perplexity_search
Direct search results with titles, URLs, snippets, and metadata for current web lookups.
←Back to Skill DirectorySuggest update
Research MCP
Perplexity MCP
Perplexity MCP gives Claude Code direct access to Perplexity\'s search and reasoning stack. Instead of juggling browser tabs, copied URLs, and manual source gathering, you can keep live search, research synthesis, and reasoning tools inside the same coding or analysis session.
For teams building research-heavy agent workflows, Perplexity MCP is useful because it separates simple search from deeper research. You can ask for a result list, a quick answer, a fully cited research pass, or a reasoning-focused analysis depending on the task.
perplexity_ask
Fast conversational answers with web context, good for everyday questions and quick checks.
perplexity_research
Deeper cited research for analysis, reports, content briefs, and topic investigation.
perplexity_reason
Reasoning-oriented flow for step-by-step analysis, planning, and harder knowledge tasks.
Installation
How to Install Perplexity MCP in Claude Code
Perplexity publishes an official Claude Code install command. The only required credential is `PERPLEXITY_API_KEY`.
Official Claude Code command
claude mcp add perplexity --env PERPLEXITY_API_KEY="your_key_here" -- npx -y @perplexity-ai/mcp-server
Manual mcpServers config
{
"mcpServers": {
"perplexity": {
"command": "npx",
"args": ["-y", "@perplexity-ai/mcp-server"],
"env": {
"PERPLEXITY_API_KEY": "your_key_here"
}
}
}
}Start with search
Use `perplexity_search` when you need a URL set, current docs, or a quick source map before any synthesis.
Escalate to ask
Use `perplexity_ask` for one fast web-backed answer when you already know the question and do not need a long memo.
Use research for briefs
Switch to `perplexity_research` when the output needs citations, a fuller narrative, or a ranked summary of findings.
Finish with reason
Use `perplexity_reason` when the hard part is judgment, prioritization, or step-by-step analysis rather than retrieval.
Execution Brief
Use this page as a rollout checklist, not just reference text.
Tool Mapping Lens
Organize Tools by Workflow Phase
Catalog-oriented pages work best when users can map discovery, evaluation, and rollout in a clear path instead of reading an undifferentiated list.
- Define the job-to-be-done first
- Group tools by stage
- Prioritize by adoption friction
Actionable Utility Module
Skill Implementation Board
Use this board for Perplexity MCP before rollout. Capture inputs, apply one decision rule, execute the checklist, and log outcome.
Input: Objective
Deliver one measurable improvement with perplexity mcp
Input: Baseline Window
20-30 minutes
Input: Fallback Window
8-12 minutes
| Decision Trigger | Action | Expected Output |
|---|---|---|
| Input: one workflow objective and release owner are defined | Run preview execution with fixed acceptance criteria. | Go or hold decision backed by repeatable evidence. |
| Input: output quality below baseline or retries increase | Limit scope, isolate root issue, and rerun controlled test. | One confirmed correction path before wider rollout. |
| Input: checks pass for two consecutive replay windows | Promote to broader traffic with fallback path active. | Stable rollout with low operational surprise. |
Execution Steps
- Record objective, owner, and stop condition.
- Execute one controlled preview run.
- Measure quality, latency, and correction burden.
- Promote only when pass criteria are stable.
Output Template
tool=perplexity mcp objective= preview_result=pass|fail primary_metric= next_step=rollout|patch|hold
What Is Perplexity MCP?
Perplexity MCP is the official MCP server for connecting AI clients to Perplexity's search and reasoning platform. In practical terms, it lets Claude Code run web-aware search and research workflows without relying on a generic browser automation loop. The server exposes distinct tools for search, quick answers, deep research, and reasoning so the model can pick the right level of depth for the job.
That distinction matters. Many teams overuse plain web search for tasks that actually require synthesis and citation. Perplexity MCP turns research into a first-class tool surface: the model can retrieve results, ask a web-backed question, produce a deeper research pass, or switch to a reasoning-oriented flow when the task becomes analytical rather than purely retrieval-based.
For agent builders, this makes Perplexity MCP one of the cleaner research plugins to add to Claude Code. It is especially strong for technical comparisons, market scans, documentation checks, content planning, and any workflow where current web context matters but you still want the model to do more than fetch snippets.
How to Calculate Better Results with perplexity mcp
Start by creating a Perplexity API key in the Perplexity console. Once you have the key, the official Claude Code install path is straightforward: add the MCP server with the documented `claude mcp add perplexity` command and pass the key through the `PERPLEXITY_API_KEY` environment variable. This keeps the credential outside your prompts and avoids hardcoding secrets into project files.
After installation, test each tool class separately. Use `perplexity_search` for a narrow current-events lookup, `perplexity_ask` for a fast factual answer, `perplexity_research` for a more complete cited analysis, and `perplexity_reason` when you want deliberate reasoning over a problem. Teams get better results when they move through those modes in order instead of treating every research task as the same depth.
Finally, wire Perplexity MCP into a broader workflow rather than using it as a standalone novelty tool. In a strong Claude Code setup, Perplexity handles current web context, memory plugins preserve reusable findings, and domain-specific MCPs like Datadog or Postgres provide operational context. That combination turns research from a one-off query into a reusable part of delivery.
Treat this page as a decision map. Build a shortlist fast, then run a focused second pass for security, ownership, and operational fit.
When a team keeps one shared selection rubric, tool adoption speeds up because evaluators stop debating criteria every time a new option appears.
Worked Examples
Example 1: Research-backed plugin evaluation
- Developer asks Claude Code to compare multiple MCP servers for a new workflow.
- Claude uses `perplexity_search` to gather current documentation and project pages, then switches to `perplexity_research` for a synthesized shortlist.
- Claude turns the findings into an internal recommendation with tradeoffs and citations.
Outcome: The team gets a current recommendation faster than manual tab-hopping and without relying on stale training data.
Example 2: Content brief creation
- Content operator needs a landing page brief on a fast-moving AI topic.
- Claude uses `perplexity_research` to map competitor claims, search intent, and cited supporting facts.
- Claude converts the research into an SEO brief and draft outline for execution.
Outcome: Research and content planning happen in one pass instead of two separate workflows.
Example 3: Fast factual verification during coding
- Engineer encounters an uncertain package change or model capability question while implementing a feature.
- Claude calls `perplexity_ask` for a quick, web-backed answer rather than leaving the terminal.
- If the answer still looks ambiguous, Claude escalates to `perplexity_reason` or `perplexity_research`.
Outcome: The team keeps momentum because research depth scales with the problem instead of defaulting to the slowest path.
Frequently Asked Questions
What is Perplexity MCP?
Perplexity MCP is Perplexity's official Model Context Protocol server. It connects AI assistants like Claude Code and Codex to Perplexity's search, research, and reasoning capabilities through a standard MCP interface.
How do I install Perplexity MCP in Claude Code?
Perplexity documents an official Claude Code install command: `claude mcp add perplexity --env PERPLEXITY_API_KEY="your_key_here" -- npx -y @perplexity-ai/mcp-server`. You can also configure it manually with an `mcpServers` block that points to `@perplexity-ai/mcp-server`.
What tools does Perplexity MCP expose?
The official docs list four main tools: `perplexity_search` for search results, `perplexity_ask` for quick search-backed answers, `perplexity_research` for deep research with citations, and `perplexity_reason` for advanced reasoning workflows.
Does Perplexity MCP work with Codex too?
Yes. Perplexity documents both Claude Code and Codex CLI setup. The Codex example uses the same `@perplexity-ai/mcp-server` package with the `PERPLEXITY_API_KEY` environment variable.
When should I use Perplexity MCP instead of a generic web search MCP?
Use Perplexity MCP when you want search and reasoning bundled together with a cleaner research workflow. A generic web-search MCP is fine for simple URL lookups, but Perplexity is stronger when you need fast synthesis, citations, and deeper analysis inside the same tool surface.
Is Perplexity MCP good for research-heavy agent workflows?
Yes. The `perplexity_research` and `perplexity_reason` tools make it especially useful for analyst workflows, technical comparisons, and content research where you want more than a plain result list.
Missing a better tool match?
Send the exact workflow you are solving and we will prioritize a new comparison or rollout guide.