Deep codebase understanding, delivered via MCP
Scope analyzes your codebase once and builds a persistent model of entities, relationships, endpoints, and conventions. Your AI agent queries it on demand.
What Scope understands about your codebase
Sync your codebase and Scope builds a structured model of your project. Not just file contents, but the relationships and patterns that matter for implementation.
Entities and relationships
Models, fields, associations, and constraints extracted from your code and schema files.
Endpoints and services
API routes, controllers, and service boundaries mapped across your entire project.
Cross-file dependencies
How files connect, what depends on what, and which changes ripple through the codebase.
Tech stack and architecture
Frameworks, libraries, patterns, and service boundaries detected automatically.
Conventions and patterns
Naming patterns, validation rules, and architecture decisions extracted from actual code.
Semantic search
Find relevant patterns, prior decisions, and related context across your entire project.
Cross-repo queries
Your agent working in one repo can pull entities, endpoints, and conventions from another project via MCP.
Under the hood: AST + schema parsing across 7+ languages, entity graph extraction, endpoint mapping, dependency analysis, and Qdrant-powered semantic search. All delivered through MCP.
How your agent accesses context
Scope exposes structured context through MCP tools that Claude Code, Cursor, and other agents call automatically as they work — with cross-repo context when your project spans multiple repositories.
Context and workflow tools
get_contextPull entities, relationships, endpoints, tech stack, and schema by scope. Your agent gets exactly the context it needs for the current task.
searchSemantic search across project patterns, conventions, and prior decisions. Find how similar problems were solved before.
start_ticketGet the next implementation ticket with full context: files to change, dependencies, acceptance criteria, and related patterns.
analyzeRun dependency analysis and impact checks. Know which files are affected before making changes.
save_learningCapture patterns, conventions, and decisions discovered during implementation. Future sessions benefit automatically.
complete_ticketMark work done and log what was learned. Scope's context model improves with every completed task.
Feature generation tools
generate_featureDescribe a feature in plain English. Scope generates implementation-ready tickets grounded in your actual codebase.
generate_bugfixDescribe a bug. Scope traces it through your code and generates a fix ticket with the right files and test criteria.
analyze impactSee which files, endpoints, and models are affected by a proposed change before any code is written.
Why this matters: Without structured context, agents burn tens of thousands of tokens exploring your codebase every session. With Scope, they spend those tokens writing code. Typical result: up to 80% fewer exploratory tokens.
Specs your AI agent can execute autonomously
Tickets include file paths, acceptance criteria, test requirements, and dependency ordering. Written for AI agents to execute without ambiguity.
Exact file references
Every spec includes the specific files to create and modify, with context about what each file does.
Testable acceptance criteria
Verifiable conditions so your agent (or your team) can confirm the work is complete.
Dependency ordering
Foundational changes ship first. Dependent work follows in the right sequence.
Convention-aware
Specs follow your project's naming patterns, architecture decisions, and coding style.
Works with your existing tools
Scope adds a context layer on top of the tools you already use. No migration, no workflow changes.