Authorization notice. Myrqen is dual-use. Only use it on software, sites and systems you own or have explicit permission to test.
Who it is forA tool evaluator or contributor who needs measured detection results instead of numbers copied from a README.
The problemDetection claims for a security scanner are hard to judge without knowing how the scanner performs on known vulnerable and safe code.
The outcomeRecall, precision, per-class results, and the safe-case false-positive rate, compared against the committed baseline.
What you need first
- A coding agent you already use, such as Claude Code or Codex
- Node.js 22 or newer and npx for the Agent Skill install
- A local project you own or are explicitly authorized to test
- Node.js 22 or newer and pnpm if you want to build from source
Steps
- Install Node.js 22 or newer and pnpm, then clone the Myrqen repository.
- Run
pnpm installin the repository root. - Run
pnpm build, then runpnpm benchmarkfor the full fixture benchmark. - Compare the output with benchmark/baseline.json and benchmark/corpus-baseline.json.
Prompt for your agent
In the stijnswapped/Myrqen repository, run `pnpm benchmark` and report recall, precision, and the safe-case false-positive rate. Expect a non-zero exit if results fall below the release gates or regress from benchmark/baseline.json. Verify that the reported numbers match the committed baseline files.
Caveat. The benchmark covers only the listed frameworks and cases in the corpus, so results do not show accuracy on every codebase.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for stijnswapped/Myrqen and "Measure the scanner's accuracy on its corpus"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer