CtrlK
BlogDocsLog inGet started
Tessl Logo

perplexity

Web search and research using Perplexity AI. Use when user says "search", "find", "look up", "ask", "research", or "what's the latest" for generic queries. NOT for library/framework docs (use Context7) or workspace questions.

94

1.62x
Quality

91%

Does it follow best practices?

Impact

99%

1.62x

Average score across 3 eval scenarios

SecuritybySnyk

Advisory

Suggest reviewing before use

SKILL.md
Quality
Evals
Security

Evaluation results

97%

44%

Perplexity Integration Helper Module

Default search parameters and Search vs Ask distinction

Criteria
Without context
With context

Default max_results

100%

100%

Default max_tokens_per_page

0%

100%

Increased limits condition: user request

50%

100%

Increased limits condition: empty results

0%

100%

Increased limits condition: complex topic

62%

100%

Increased max_results cap

33%

100%

Increased max_tokens_per_page cap

0%

100%

Search for URLs/results

100%

100%

Ask for conversational answers

100%

100%

perplexity_research absent

100%

70%

Ask messages format

0%

100%

100%

38%

Developer Assistant Tool Routing Guide

Tool priority chain and query routing

Criteria
Without context
With context

Context7 for library docs

80%

100%

Graphite MCP for gt CLI

100%

100%

Nx MCP for workspace

100%

100%

Perplexity Search for generic searches

75%

100%

Perplexity Ask for conversational

0%

100%

Researcher agent for deep research

0%

100%

WebSearch as last resort

0%

100%

Prohibited tool named

0%

100%

Library doc example routes to Context7

100%

100%

gt CLI example routes to Graphite

100%

100%

Priority ordering correct

100%

100%

100%

30%

Audit and Fix an AI Assistant Tool Configuration

Prohibited tool avoidance and trigger keyword routing

Criteria
Without context
With context

Identifies perplexity_research violation

100%

100%

Fixes perplexity_research

100%

100%

Routes library docs to Context7

0%

100%

Identifies library doc misrouting

50%

100%

Routes gt CLI to Graphite MCP

100%

100%

Identifies gt CLI misrouting

100%

100%

Fixes default max_results

0%

100%

Fixes default max_tokens_per_page

100%

100%

Identifies default parameter violations

100%

100%

Deep research replacement named

50%

100%

Repository
softaworks/agent-toolkit
Evaluated
Agent
Claude Code
Model
Claude Sonnet 4.6

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.