Skip to main content
Glama
Sign In
deenesjakoruzh
Jercik

Perplexity Agent MCP

by Jercik
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Remote

Perplexity Agent MCP

Source: https://code.j4k.dev/j4k-oss/perplexity-agent-mcp. github.com/Jercik/perplexity-agent-mcp is a read-only mirror; open issues and pull requests on the Forge.

An opinionated, coding-specialized, Model Context Protocol server that integrates Perplexity AI's Sonar models into AI coding assistants like Claude Code or Codex CLI. It's deliberately lean and adds very few tokens to your context. It covers quick API fact-checks and deeper research; How it works describes the tools.

Setup

The package is published only as @j4k-oss/perplexity-agent-mcp on the code.j4k.dev registry. The unscoped perplexity-agent-mcp package on npmjs is deprecated and gets no updates.

Always map the @j4k-oss scope to the code.j4k.dev registry: without a mapping, npm looks on npmjs, finds nothing, and fails with a 404. To do that once for every command, add this line to ~/.npmrc:

@j4k-oss:registry=https://code.j4k.dev/api/packages/j4k-oss/npm/

The inline flag in the commands below is enough for a single npx run. Keep the ~/.npmrc line for the global install: npm update -g and npm outdated -g read the registry from your configuration, so without the line they look the scoped name up on npmjs.org. Install the package globally:

npm install -g --@j4k-oss:registry=https://code.j4k.dev/api/packages/j4k-oss/npm/ @j4k-oss/perplexity-agent-mcp

If you installed the unscoped package globally, run npm uninstall -g perplexity-agent-mcp (or pnpm remove -g perplexity-agent-mcp) first: both packages provide the perplexity-agent-mcp command, npm refuses to overwrite it, and pnpm keeps running the old copy. To check the latest release without installing it:

npx -y --@j4k-oss:registry=https://code.j4k.dev/api/packages/j4k-oss/npm/ @j4k-oss/perplexity-agent-mcp@latest --version

In a pnpm project, map the scope to its registry and exempt it from minimumReleaseAge in pnpm-workspace.yaml, then add the package:

registries:
  "@j4k-oss": https://code.j4k.dev/api/packages/j4k-oss/npm/
minimumReleaseAgeExclude:
  - "@j4k-oss/*"
pnpm add -D @j4k-oss/perplexity-agent-mcp

The package installs the perplexity-agent-mcp command and reports the MCP server name perplexity-agent-mcp.

Then add it to your Claude config file (~/.claude.json or ~/.config/claude/config.json):

"mcpServers": {
  "perplexity": {
    "command": "perplexity-agent-mcp",
    "env": {
      "PERPLEXITY_API_KEY": "your-api-key-here"
    }
  }
}

Restart Claude Code after saving, and you're ready to go.

How it works

The server exposes only two tools with concise descriptions (~1.3K tokens), keeping your context window lean and efficient:

  • lookup: Gets quick facts from documentation (like API syntax or config keys)

  • answer: Does deeper research to compare options and make recommendations

Under the hood, both tools are opinionated wrappers that call Perplexity API with coding-specialized system prompts. The lookup tool uses a compact fact‑extraction prompt focused on code/docs facts (see LOOKUP_SYSTEM_PROMPT in src/prompts.ts). The answer tool uses a technical decision/analysis prompt tailored for migrations and architecture choices (see ANSWER_SYSTEM_PROMPT in src/prompts.ts). Each tool calls a fixed Perplexity model chosen for its task, and callers cannot select another. The models are set in registerLookupTool (src/tools/lookup.ts) and registerAnswerTool (src/tools/answer.ts).

Example AI agent prompt that will trigger AI agent to use this MCP and give you good results:

Please update the "foo" dependency to the latest version. Use Perplexity for the migration guide.

Requirements

  • Node.js v24.15.0 or newer

  • A Perplexity API key

Available Tools

2 tools
answerA

Research a question, compare options, and recommend a path (backed by sources). Use for library choices, architecture trade-offs, migrations, complex debugging, and performance decisions. Returns a concise recommendation, a brief why, and short how-to steps.

NameRequiredDescriptionDefault
questionYesThe decision or problem to answer

TDQS

A3.7/5.0
Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations are provided, so the description carries the full burden of behavioral disclosure. It describes the tool's behavior as involving research, comparison, and recommendation backed by sources, and specifies the return format ('concise recommendation, a brief why, and short how-to steps'). However, it doesn't cover other important aspects like whether it requires external data access, potential rate limits, or error handling. With no annotations, this is a moderate but incomplete disclosure.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is appropriately sized and front-loaded, with three sentences that each add value: the first states the core function, the second provides usage examples, and the third specifies the return format. There is no wasted text, and it efficiently conveys key information without redundancy.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the complexity (a research and recommendation tool with no annotations and no output schema), the description is moderately complete. It explains the tool's purpose, usage, and return format, but lacks details on behavioral traits like data sources, limitations, or error cases. Without an output schema, it does describe the return values, which helps, but overall it could be more comprehensive for such a tool.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The input schema has 100% description coverage, with the 'question' parameter documented as 'The decision or problem to answer.' The description doesn't add any further details about parameters beyond what the schema provides. According to the rules, when schema_description_coverage is high (>80%), the baseline is 3 even with no param info in the description, which applies here.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool's purpose: 'Research a question, compare options, and recommend a path (backed by sources).' It specifies the verb ('research, compare, recommend') and resource ('question'), making the function evident. However, it doesn't explicitly distinguish this from the sibling tool 'lookup', which might be a similar search or query function, so it lacks sibling differentiation for a perfect score.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description provides clear context for when to use this tool: 'Use for library choices, architecture trade-offs, migrations, complex debugging, and performance decisions.' This gives specific examples of applicable scenarios. However, it doesn't mention when not to use it or explicitly compare it to alternatives like the sibling tool 'lookup', so it falls short of the highest score.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

lookupA

Fetch precise, source-backed facts from official sources. Use for API syntax/params, config keys/defaults, CLI flags, runtime compatibility, and package metadata (types, ESM/CJS, side-effects). Returns short, factual answers. No recommendations or comparisons.

NameRequiredDescriptionDefault
queryYesWhat fact to look up from docs

TDQS

A4.4/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations provided, the description carries the full burden and does well by disclosing key behavioral traits: it fetches from official sources, returns short factual answers, and avoids recommendations/comparisons. It doesn't mention rate limits, authentication needs, or error handling, but provides solid operational context.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is perfectly concise with three sentences that each earn their place: first states purpose, second specifies use cases, third clarifies return behavior and exclusions. No wasted words, front-loaded with core functionality.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the single parameter with full schema coverage and no output schema, the description provides excellent context about what the tool does, when to use it, and what to expect. It could benefit from mentioning response format or error cases, but covers the essential operational context well.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100% with the single parameter 'query' well-documented in the schema. The description adds minimal value beyond the schema by implying the query should be factual lookups, but doesn't provide additional syntax or format details. Baseline 3 is appropriate when schema does the heavy lifting.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool's purpose with specific verbs ('fetch precise, source-backed facts') and resources ('from official sources'), and distinguishes it from its sibling 'answer' by specifying it returns short factual answers without recommendations or comparisons.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description provides explicit guidance on when to use this tool ('for API syntax/params, config keys/defaults, CLI flags, runtime compatibility, and package metadata') and when not to use it ('No recommendations or comparisons'), clearly differentiating it from the 'answer' sibling tool.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

  1. 2 tool updatesv1.0.0
    • First observedanswer
    • First observedlookup
Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

A3.9/5.0

Scored across 2 tools

Disambiguation5/5

The two tools have clearly distinct purposes: 'answer' is for research, comparison, and recommendations on complex decisions, while 'lookup' is for fetching precise, factual information from official sources. There is no overlap in functionality, as one provides recommendations and the other provides facts without recommendations.

Naming Consistency5/5

Both tool names follow a consistent, simple verb-based pattern ('answer' and 'lookup') that clearly indicates their actions. There are no deviations in naming conventions, making them predictable and easy to understand.

Tool Count3/5

With only two tools, the set feels thin for a server named 'Perplexity Agent MCP', which suggests a broader scope for handling queries and research. While the tools cover distinct areas, more operations (e.g., for follow-up queries or deeper analysis) might be expected to fully support agent workflows.

Completeness3/5

The tools cover two key areas—complex decision-making and factual lookups—but there are notable gaps. For instance, there are no tools for updating or managing research sessions, handling multi-step interactions, or integrating with external data sources beyond the described lookups, which could limit agent effectiveness in extended tasks.

Maintenance

ActivityActive
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers

  • Perplexity MCP Server

    sengokudaikon
    A
    license
    Not graded
    quality
    D
    maintenance
    Provides access to Perplexity AI models through two tools: ask\_perplexity for expert programming assistance and chat\_perplexity for maintaining ongoing conversations with context preservation.
    MIT

  • MCP Perplexity Server

    samwang0723
    A
    license
    Not graded
    quality
    D
    maintenance
    Provides AI-powered search, research, and reasoning capabilities through integration with Perplexity.ai, offering three specialized tools: general conversational AI, deep research with citations, and advanced reasoning.
    18 npm
    2
    MIT

  • Perplexity MCP Server

    Pantainos-Toolkit
    F
    license
    Not graded
    quality
    D
    maintenance
    Integrates with Perplexity's API to provide web search and AI-powered answers with citations. Offers a three-tier research workflow: search for sources, ask for grounded AI answers, and ask_more for deeper analysis using advanced models.
    -

Appeared in Searches