SpecThis

Getting Started with SpecThis

Create structured plans for AI coding agents, monitor code health, and integrate with your development workflow

What is SpecThis?

SpecThis is a development platform that helps engineering teams work more effectively with AI coding agents. It provides three core capabilities:

Structured Plans

Create detailed specifications with work items and file-level changes that AI coding agents can consume directly. Plans give agents the context they need to produce accurate, consistent code on the first pass.

Code Intelligence

Automated scanning analyzes your codebase across 7 metrics including complexity, file size, and symbol count. Configurable alerts notify you when code exceeds your thresholds, and a dashboard tracks trends over time.

MCP Integration

A Model Context Protocol (MCP) server exposes 16 tools that let AI agents read and write plans, manage work items and changes, and link repositories — all from within your existing AI workflow in Claude Code, Cursor, or Codex.

Quick Start

Get up and running in six steps:

1

Sign Up

Create your account at app.specthis.ai. Every account starts on the free Hobby tier — no credit card required.

2

Connect GitHub

Install the SpecThis GitHub App on your repositories from your organization settings. This enables code scanning and repository-aware planning.

3

Install the MCP Server

Add the SpecThis MCP server to your AI agent (Claude Code, Cursor, or Codex). See the MCP Setup guide for step-by-step instructions.

4

Create Your First Plan

Ask your AI agent to create a plan for a feature you are working on. The agent uses MCP tools to create the plan, add work items, and specify file changes — giving you a structured specification you can review before implementation begins.

5

Run a Scan

Run npx spec-this scan from your repository to analyze code complexity. The CLI uploads results to your dashboard automatically. See the CLI setup guide for details.

6

View Your Dashboard

Your dashboard shows health scores, metric trends with sparklines, and active alerts. Use it to monitor code quality over time and catch regressions early.

Core Features

Plans

Plans are structured specifications that describe what needs to be built. Each plan contains:

  • Work items — individual tasks with descriptions, acceptance criteria, and status tracking
  • File changes — specific files to create, modify, or delete with detailed instructions
  • Plan structure — hierarchical organization of work items for complex features
  • Repository links — connect plans to GitHub repositories for context-aware development

Code Intelligence

The scanner analyzes your codebase and tracks 7 metrics:

  • complexity_avg — average cognitive complexity across all functions
  • complexity_max — maximum single-function complexity score
  • complexity_p95 — 95th percentile complexity (top 5% most complex)
  • file_count — total number of source files scanned
  • total_loc — total logical lines of code across all functions
  • symbol_count — number of symbols (functions, classes, methods) found
  • alert_count — number of active code alerts by severity

See the Code Metrics overview for detailed explanations of each metric.

MCP Tools

The MCP server provides 16 tools organized into five groups:

Plans

  • read_plans
  • upsert_plan
  • delete_plan

Work Items

  • read_work_items
  • upsert_work_item
  • delete_work_item

Changes

  • read_changes
  • upsert_change
  • delete_change

Repositories

  • read_repositories
  • read_plan_repositories
  • link_plan_repository
  • unlink_plan_repository

Plan Structure

  • read_plan_structure
  • upsert_plan_structure_item
  • delete_plan_structure_item

Dashboard

The dashboard provides a real-time view of your codebase health. It includes repository health scores, metric sparklines showing trends, active alert counts by severity, and historical comparisons so you can see whether code quality is improving or declining over time.

Plan Tiers

Hobby

$0 / forever

  • Unlimited plans and work items
  • Code scanning via CLI or GitHub Actions
  • 10 scans per day
  • 250 MB / 5,000 file repo limit
  • 1 month raw data retention
  • 7 code metrics with alerts
  • Dashboard with basic trends

Pro

$49 / month

  • Everything in Hobby, plus:
  • Managed scanning via Trigger.dev (we pay compute)
  • 100 scans per day
  • 1 GB / 50,000 file repo limit
  • 1 year raw + weekly aggregates
  • Monthly aggregates retained forever
  • Advanced dashboard with full trend history

Next Steps

Set Up MCP Integration →Explore Code Metrics →Install the CLI →