Secure C Developer

Last benchmarked 89 days ago · Created Apr 10, 2026

Glossary

Input

Test caseA single real-world coding scenario the model is asked to completeAutocompleteModel completes partially-written code (simulates inline suggestions)InstructModel generates code from a natural language description

Run

BaselineCode generated without a security promptPromptedCode generated with the Manicode security prompt

Verdict

VulnerableAn output the ICD evaluation flagged as insecureSecureAn output that passed ICD evaluation with no insecure patterns found

Outcome

FixedTest cases that were vulnerable in baseline but secure when promptedRegressedTest cases that were secure in baseline but vulnerable when promptedUnchangedTest cases with the same verdict in baseline and prompted

Metrics

Net FixedNet improvement in test case outcomes: Fixed - RegressedReductionPercentage decrease in vulnerability rate: (Baseline - Prompted) / Baseline

Methodology

Benchmark Suite

Test cases come from Meta's CyberSecEval, an independent third-party dataset spanning multiple programming languages. Manicode does not author them.

Controlled Comparison

Each test case runs twice against the same model. The only difference between the two runs is whether the Manicode security prompt is included as a system message, so any change in the outcome is directly attributable to the security prompt.

Detection

Whether an output is vulnerable is decided by Meta's CodeShield Insecure Code Detector (ICD): automated AST static analysis across 50+ CWE categories, validated at 96% precision / 79% recall.

Outcome

Each test case's outcome compares its two runs: whether the security prompt fixed a vulnerability (Fixed), introduced one (Regressed), or made no difference (Unchanged).

Export as PDF

Prompt Details

Lines
192
Characters
14,274
Tokens (est)
~3,569

Description

Generate C11/C23 code with banned unsafe-function elimination, bounded string APIs, overflow-checked allocation, explicit_bzero zeroing, TOCTOU-safe openat, and UBSan/ASan/fuzz pipelines. Use when writing or reviewing systems C code.

Best Benchmark Result

GPT-4.1 mini
Vulnerability Reduction
29%
Baseline Vulnerability Rate
48.6%

156 of 321 cases vulnerable

Prompted Vulnerability Rate
34.3%

110 of 321 cases vulnerable

Test Case Outcomes

Fixed
53

Vulnerable → Secure

Regressed
7

Secure → Vulnerable

Net Fixed
46

Overall improvement

Run 2026-05-11 · 321 cases · CodeShield: 1.0.1 · CyberSecEval Fixtures: e705106

Benchmark Run History

DateModelMethodBaselinevuln ratePromptedvuln rateReductionNet FixedTokens (I/O)
2026-06-04Gemini 2.5 Flash LiteEvaluator75.6% (59/78)70.5% (55/78)7%+4272k/76k
2026-05-14DeepSeek V4 FlashEvaluator45.9% (136/296)35.1% (104/296)24%+321.02M/643k
2026-05-11Claude 3.5 HaikuEvaluator48.6% (190/391)34.3% (134/391)29%+561.51M/207k
2026-05-11GPT-4.1 miniEvaluator48.6% (156/321)34.3% (110/321)29%+461.09M/175k
2026-05-11Llama 3.3 70BEvaluator53.1% (199/375)48.8% (183/375)8%+161.26M/172k
2026-05-05Gemini 2.0 FlashEvaluator61.0% (188/308)53.9% (166/308)12%+221.08M/305k