wycheproof
by trailofbitsThe wycheproof skill helps validate cryptographic implementations with Wycheproof test vectors, focusing on known attacks, edge cases, and pass/fail decisions for Security Audit workflows. Use it to inspect AES-GCM, ECDSA, ECDH, RSA, and related primitives with less guesswork than a generic crypto prompt.
This skill scores 78/100, which means it is a solid listing candidate for directory users. The repository provides enough real workflow content to help an agent recognize when to use Wycheproof and apply it to crypto-testing tasks with less guesswork than a generic prompt, though it lacks some adoption aids like an install command or supporting files.
- Clear trigger and scope for cryptographic validation: the frontmatter says to use it for testing crypto code for known attacks and edge cases.
- Substantial operational content: the body is large and structured with many headings, workflow sections, and code fences, giving an agent enough material to follow.
- Concrete domain value: it explains key concepts like test vectors, test groups, and result flags, which improves agent leverage for implementation testing.
- No install command or companion scripts/resources, so users may need to wire invocation and execution steps themselves.
- The description is concise and support-file sparse, so some environment-specific setup and integration details may still require manual interpretation.
Overview of wycheproof skill
What wycheproof is for
The wycheproof skill helps you use Wycheproof test vectors to validate cryptographic implementations against known edge cases and attack patterns. It is best for a Security Audit workflow when you need to check whether a library, service, or product accepts the right inputs, rejects the wrong ones, and fails safely on ambiguous cases.
Who should install it
Install the wycheproof skill if you audit crypto code, maintain a security-sensitive application, or need a repeatable way to test primitives like AES-GCM, ECDSA, ECDH, or RSA. It is especially useful when a quick prompt is not enough because the failure mode is subtle: the code may “work” on happy-path examples while still being vulnerable.
What makes it different
Wycheproof is not a general crypto tutorial. The value of the wycheproof skill is that it focuses on known-bad inputs, structured test groups, and pass/fail/acceptable outcomes. That makes it more useful than a generic “test my crypto” prompt when you need decision-quality results for a review or fix.
How to Use wycheproof skill
Install and inspect the skill
Use the wycheproof install flow from your skills manager, then open SKILL.md first. In this repo, SKILL.md is the only support file, so there is no separate script or rules layer to learn. That means your main job is to extract the workflow, test categories, and constraints from the skill body itself.
Turn a rough goal into a useful prompt
The wycheproof usage works best when you provide the crypto primitive, implementation language, and testing goal up front. A weak request is: “Check my crypto code.” A stronger request is: “Use the wycheproof skill to test our Java ECDSA verifier against malformed signatures and boundary cases; report which vectors should pass, fail, or be treated as acceptable by our policy.”
Start with the right inputs
For best results, include:
- the algorithm or protocol
- the language or library
- what counts as success in your environment
- whether you want regression testing, audit support, or triage of a failing case
- any constraints such as FIPS mode, legacy compatibility, or platform limits
These details matter because wycheproof output changes depending on whether you are validating strict rejection, compatibility behavior, or a known exception list.
Read the workflow in order
A practical wycheproof guide should be read as: background, key concepts, when to use, then the testing workflow sections. If you are using it for a Security Audit, pay special attention to how the skill distinguishes valid vectors from invalid or acceptable ones, because that classification is the main thing that prevents false conclusions.
wycheproof skill FAQ
Is wycheproof only for security audits?
No. The wycheproof skill is valuable for audits, but it is also useful during implementation and regression testing. If you ship cryptography, this skill helps you catch issues before they become audit findings.
Do I need to be a crypto expert first?
No, but you do need enough context to name the primitive and describe the expected behavior. If you cannot say whether a vector should pass, fail, or be accepted under a compatibility rule, the output will be less actionable.
How is this different from a normal prompt?
A normal prompt may produce a generic checklist. The wycheproof skill is better when you need structured test-vector reasoning and edge-case coverage. It reduces guesswork by anchoring the task to known attack patterns instead of broad advice.
When should I not use it?
Do not use wycheproof if you are trying to design a crypto protocol from scratch or explain cryptography at a high level. It is for validation and testing, not for theoretical teaching or product architecture.
How to Improve wycheproof skill
Give the skill the exact crypto target
The biggest quality boost comes from naming the primitive, implementation, and failure mode. For example, “RSA-PSS verifier in Python cryptography, failing on certain salt lengths” is far better than “my signature code is broken.” The more exact the target, the easier it is to map to the right Wycheproof vectors.
State your acceptance policy early
One common failure mode in wycheproof work is mixing security correctness with product compatibility. Say whether you want strict rejection of invalid vectors, compatibility with legacy inputs, or a documented allowlist. That distinction changes the result and prevents noisy back-and-forth after the first pass.
Iterate on the vector class, not just the bug
If the first output finds a failure, refine the next request by asking for adjacent cases: nearby key sizes, malformed encodings, truncated inputs, or boundary values from the same test group. That is usually more useful than asking for a broad rerun, because Wycheproof’s strength is coverage of families of edge cases.
Use the output to drive regression tests
After you confirm a finding, convert the failing case into a permanent test in your own suite. The wycheproof skill is most valuable when it produces a durable security audit trail: which vector failed, why it failed, and what condition should be enforced on future releases.