stryker-mcp-reporter v1.13.0 adds an ESLint hook and lets AI agents run Stryker mutation tests to chase 100% mutation score. The claim lacks reproducible proof.
stryker-mcp-reporter v1.13.0 ships an ESLint hook for McCabe complexity, letting Cursor and Claude Desktop agents chase a verified 100% Mutation Score. The open-source MCP server turns Stryker's mutation testing into an autonomous AI loop.
Key facts
- v1.13.0 adds ESLint hook for McCabe complexity and max-lines
- Supports Antigravity, Cursor, Claude Desktop, Roo Code, Cline
- Config uses single JSON block with npx command
- Open-source on GitHub under MIT license
- Claims verified 100% Mutation Score via MCP loop
Standard code coverage measures execution, not test strength. AI coding assistants generate hundreds of lines of unit tests, but often fall into the "happy path bias" — testing what works instead of what breaks. The project's announcement positions v1.13.0 as the fix for that gap.
The new release is thin on features: one commit adds an ESLint hook that enforces McCabe complexity and max-lines limits (commit 330a951). The rest of the value is the MCP integration itself, which lets AI pair programmers — Antigravity, Cursor, Claude Desktop, Roo Code, Cline — execute mutation tests, inspect survived mutants via Model Context Protocol, and write edge-case tests to close the gap.
Key Takeaways
- stryker-mcp-reporter v1.13.0 adds an ESLint hook and lets AI agents run Stryker mutation tests to chase 100% mutation score.
- The claim lacks reproducible proof.
Why Mutation Score Beats Coverage
Mutation testing mutates your source code — flipping operators, deleting statements — and checks whether your tests catch the change. A 100% Mutation Score means every injected fault was detected. That's a materially stronger signal than line coverage, which can hit 100% while your tests assert nothing meaningful. The tradeoff has always been cost: mutation testing is slow and noisy. Stryker, the underlying framework, mitigates this with incremental runs, but it still demands more compute than a standard test suite.
The MCP angle is the structural shift. Instead of a developer reading a mutation report and writing tests by hand, the AI agent does the loop: run Stryker, read the survived mutants, write a targeted test, re-run. The screenshots in the post show an inline code diff of a survived mutant and the resulting report. The author claims a "verified 100% Mutation Score" is achievable through this loop.
The Verification Gap
The claim deserves scrutiny. The post shows runtime screenshots, not a reproducible benchmark. There's no CI pipeline published with the repo proving the 100% score on a non-trivial codebase. The author doesn't disclose the test suite size, the number of mutants, or the Stryker configuration used. For a tool whose entire pitch is test strength, the absence of a public, runnable proof is a notable omission.
That said, the underlying problem is real. The same ecosystem that produced this tool also produced a wave of unverified MCP servers — the project's own documentation notes that most MCP servers ship without any verification of protocol compliance, schema integrity, or cross-client compatibility. stryker-mcp-reporter is at least honest about its scope: it's a reporter that surfaces mutation results to the agent, not a verification framework for MCP itself.
Setup and Compatibility
Connecting the tool takes a single JSON block in your MCP config:
{
"mcpServers": {
"stryker-mutation-testing": {
"command": "npx",
"args": ["-y", "--silent", "stryker-mcp-reporter"]
}
}
}
Windows users swap in cmd.exe with /c as the first arg. The project is open-source on GitHub under the MIT license.
The Bottom Line
For teams already running Stryker, this tool lowers the friction of closing mutation gaps. For teams treating AI-generated tests as finished because coverage is high, it's a useful corrective — if the agent actually acts on the survived mutants. The 100% claim is a marketing headline until the repo ships a reproducible proof. Watch for whether the author publishes a CI workflow with a public mutant count.
What to watch
Watch for a reproducible CI workflow in the GitHub repo that publishes a real mutant count and mutation score on a non-trivial codebase. If the author ships that, the 100% claim becomes verifiable. Also track whether Stryker's own team adopts MCP natively, which would subsume this project.
Source: dev.to
[Updated 10 Aug via devto_mcp]
The project has already advanced past v1.13.0: v1.14.0 added hooks for ADR creation and evaluation (commit 8b225dd), and v1.15.0 implements a Hybrid Auto-Remediation Profiling pipeline (commit 4c8825e). The new pipeline reportedly automates profiling of survived mutants to guide remediation, though no reproducible CI proof accompanies the claim. [per dev.to]
Originally published on gentic.news
Top comments (0)