gRPC and Protobuf context engine for AI coding agents
ProtoMap is a context engine and model harness for gRPC and Protobuf, built by HoundDog.ai. It scans source code to map every service, message, and field across an organization, builds a graph updated on every commit, and serves that context to AI coding agents over MCP and to humans through a model harness. According to vendor benchmarks, agents answering with ProtoMap use 13× fewer tokens, answer 7.7× faster, and cost 6.4× less than agents searching raw code.
Scans source code and derives a deterministic graph of gRPC services, methods, messages, and fields — no instruction or memory files to write or maintain.
Agents such as Claude Code, Codex, Antigravity, VS Code, and OpenCode query the API graph over the Model Context Protocol, answering questions like which services call a method and where.
Runs protomap scan on the local machine; a million lines of code analyzed in seconds, no GPU required, and no AI tokens consumed for the mapping itself.
The graph updates on every merge to main (via CI scans on pull requests) and in real time for local code changes.
Bring your own AI model (Anthropic, OpenAI, Grok, Gemini, AWS Bedrock, Microsoft Foundry, TypeSafe.ai) to prune false positives from the call graph.
Query the graph directly for blast radius, field lineage, call graphs, dependencies, service hotspots, dead gRPC methods, and code owners.
Supports Go, Java, Kotlin, C#, Python, TypeScript, JavaScript, Rust, and Swift, across monorepos and multi-repo setups.
Runs on-premises with Docker and PostgreSQL, or fully managed in the cloud; both options are SOC 2 compliant with SSO, RBAC, and audit logs.
Run protomap scan to analyze your gRPC and Protobuf code — a million lines of code in seconds across monorepos or multiple repos.
Entities (repos, services, methods, messages, fields) and their connections form a graph, updated on every merge to main and in real time for local code changes.
Bring your own AI model to prune false positives from the derived call graph.
Coding agents query the graph over the ProtoMap MCP; humans query it through the model harness.
ProtoMap has a free local scanner tier and a custom-priced Enterprise tier for organization-wide, continuously updated context. Pricing for Enterprise is not published; you must contact sales.
Free
Run it on your machine and plug it into any MCP-compatible AI agent. Scans whatever code is available locally, provides local API graphs for documentation, supports gRPC (GraphQL and REST coming soon).
Custom (contact us)
Cloud or on-premises deployment with centralized, continuously updated context across all code repos and cross-service relationships. Both options are SOC 2 compliant with SSO, RBAC, and audit logs.
It scans your source code to map every gRPC service, protobuf message, and field across your organization's repositories, builds a continuously updated graph of that context, and serves it to AI coding agents over MCP and to humans through a model harness.
Yes — the Local Scanner tier is free to run on your machine and plug into any MCP-compatible AI agent. An Enterprise tier for centralized, org-wide context is priced custom via contact with sales.
Currently gRPC and Protobuf. GraphQL and REST support is listed as coming soon.
Go, Java, Kotlin, C#, Python, TypeScript, JavaScript, Rust, and Swift.
Any MCP-compatible agent, including Claude Code, Codex, Antigravity, VS Code, and OpenCode.
Both. The local scanner runs on your machine; Enterprise offers a fully managed SaaS cloud platform or an on-premises deployment that runs entirely inside your network, with Docker and PostgreSQL. Both options are SOC 2 compliant.
It is free on GitHub, and self-hosted installation is done by cloning the GitHub repository and running an install script. The page does not state a specific open-source license.
Pricing is not published; you need to book a demo or contact sales for a quote.