tests: mutation-harden rag_guard/guard.py — 25/25 surviving mutants killed (generated by crucible)

[Jott2121]

What

Adds 26 generated tests that kill all 25 mutants the existing suite missed in rag_guard/guard.py, plus the [tool.mutmut] scope config that proved the setup.

Baseline: mutation testing injected 71 defects into guard.py; the existing suite killed 46 and 25 survived — real gaps, each one an injected bug no test caught. After this PR: 71/71 killed.

How these were made

Generated by crucible — an adversarial test-hardening loop (Tester agent → mutmut → Critic agent handed the named survivors). Every verdict here is mechanical: pytest kills the mutant or it survives; no model judged model output. Receipts (per-round tokens, kills, cost, prompt hashes): run 20260712T050833Z-rag-guard-harden. Tester killed 23, Critic killed the last 2. $0 metered (Claude subscription), shadow-priced $1.63.

The tests were reviewed as part of crucible's owner-approved acceptance gate before this PR was opened.

Verification

• Full rag-guard suite green with these tests on pristine code (88 passed).
• Each test targets named survivors; kill receipts in the run dir above.

🤖 Generated with Claude Code

[github-actions]

Coverage report

Click to see where and how coverage changed

File	Statements	Missing	Coverage	Coverage (new stmts)	Lines missing
rag_guard
guard.py
Project Total

This report was generated by python-coverage-comment-action