zephex
⌘K
Get StartedPricingMCP ToolsDocs
←Back×Sign in
Get StartedPricingMCP ToolsDocs

Documentation

Start with the endpoint and API key flow, then branch into editor setup or tool reference.

GET STARTED

What is MCP?IntroductionQuickstartAPI ReferenceUniversal RequirementsZephex vs Local MCPBest PracticesToken EfficiencyCLAUDE.md TemplateAGENTS.md TemplateMCP EcosystemMarkdown Access

EDITORS

OpenCodeOAuthVS CodeClaude CodeCursorWindsurfJetBrainsKiro CLIOAuthZedOAuthGemini CLIOAuthClineOAuthGooseCodex CLIOAuthFactory AIOAuth

OPERATING SYSTEMS

macOSWindowsLinux

TOOLS

get_project_contextread_codefind_codecheck_packageaudit_packageexplain_architectureZephex_dev_infoscope_taskaudit_headersthinking

SUPPORT

PlansPro & Max GuideUsage & AnalyticsConnection IssuesRate LimitsBillingSecurityFAQChangelog

thinking

Creates a persistent reasoning session so you can track what was checked, what was concluded, and what to do next.

DESCRIPTION

thinking is not a code reader. It is a stateful investigation layer for debugging, high-risk changes, and long sessions where it is easy to lose track of the work.

WHEN TO USE

GOOD FIT

  • During complex debugging across several files or systems.
  • Before high-risk changes such as auth or billing refactors.
  • When an initial fix failed and you need a structured re-evaluation.
  • When you want a written investigation trail instead of ad hoc chat history.

AVOID IT WHEN

  • The task is simple and scope_task plus read_code is enough.
  • You only need code or search output, not investigation state.
  • You do not intend to carry the reasoning across multiple steps.
PARAMETERS
NameTypeRequiredDescription
operation"generate" | "cleanup"YesCreate a new reasoning session or clean up old ones.
task_descriptionstringNoProblem statement when creating a new session.
project_pathstringNoAbsolute project path to anchor the investigation.
DETAILS

PROJECT PATH

When to include project_path

Add project_path when the reasoning session is tied to one repo. That gives downstream tool recommendations a concrete workspace anchor. If the problem is cross-system or mostly conceptual, you can omit it.

EXAMPLE
tools-call.json
{  "jsonrpc": "2.0",  "id": 10,  "method": "tools/call",  "params": {    "name": "thinking",    "arguments": {      "operation": "generate",      "task_description": "Investigate intermittent unauthorized responses on the MCP endpoint",      "project_path": "/Users/name/projects/zephex"    }  }}
OUTPUT
response.txt
Session createdVerified state: unauthorized responses spike after API key rotationInvestigation plan: inspect key validation, cache invalidation, retry handlingRecommended next tool: read_code

TOKEN EFFICIENCY

A thinking session usually starts around 300 to 900 tokens. That overhead is worth it when it prevents repeated dead-end reads and forgotten conclusions across a long debugging session.