What ripwire MCP server does
The ripwire MCP server helps coding agents work from a repository map instead of starting with broad file reads. Its analysis ranks relevant code relationships and signatures, showing what may need changing, which callers could be affected, and which tests reach a target. It also supports targeted recall from planning documents and other repository material.
Before an edit, ripwire can help identify callers, fields read or written, ambiguous name resolution, forgotten co-changes, and likely test coverage. After an edit, it can inspect the change for contract compatibility, estimate blast radius, identify tests to run, and report quality categories that became worse. The documented quality checks include issues such as duplication and other regressions, rather than a general claim that the code is correct.
How it works
The project is implemented as a C++23 command-line tool and MCP server with no runtime dependencies. Its repository analysis produces deterministic, ranked results intended to give an agent a compact map of relevant context. Signatures are used to reduce the amount of source text an agent must consume, while the tool labels guesses and publishes information about losses or uncertainty.
Typical use starts with repository orientation, such as asking an agent to use ripwire on the current repository. A caller query can reveal whether a symbol has production callers. An edit check can compare a changed public symbol with its call sites, while a quality-delta check reports regressions introduced by the work. Persistent notes can be attached to symbols so later lookups surface project-specific gotchas across separate agent contexts.
The tool is intended as an aid to investigation and verification, not as a replacement for compilation, tests, or purpose-built measurements. Its documentation describes a known arity-heuristic false positive for defaulted trailing parameters; those results should be confirmed by compiling. Broad conceptual searches may still be better served by ordinary grep and direct reading.
Setup and configuration
The repository provides an installation path intended to be used before coding sessions, and the installation teaches an agent when to invoke the available commands. The supplied material does not specify the installation command, package format, MCP transport, client configuration, or runtime environment variables, so those details should be taken from the project’s current release documentation.
The implementation targets C++23 and advertises no runtime dependencies. It can be used as a standalone CLI or through its MCP server interface. Because the provided material does not list client-specific setup, compatibility should be checked against the client’s support for the server transport used by the selected release.