CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

rust-determinism-engineering

Use when writing deterministic tests, golden/snapshot suites, or any Rust whose output must be byte-stable across runs, machines, and parallel test execution — or when a test is flaky. In greenfield there is no reference oracle, so your own goldens ARE the spec and a flaky golden is a spec that lies. Covers the single-entropy-fixture funnel, stable hashing/ordering, golden checksum-gating, the four flaky-test causes, seeded randomness, and cross-architecture drift. testing-strategy says WHERE determinism is injected; this is the Rust-deep HOW.

Quality

83%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

SKILL.md
Quality
Evals
Security

Quality

Content

75%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A well-engineered, experience-backed rulebook: every rule carries concrete Rust-level mechanics and field provenance, and validation checkpoints (checksum-gating, proof-obligation re-blessing, warmup-before-measure) are explicit. The gaps are the absence of any complete code skeleton for the central entropy fixture and a motivational intro that could be tightened.

Suggestions

Add a minimal executable skeleton of the single entropy fixture (clock counter, seeded PRNG, BTreeMap env, seed-derived run_id in one struct) — the skill's core concept currently has no code shape.

Compress the three-paragraph intro and "Field origin" paragraph into 2–3 sentences; the rule sections already convey severity.

DimensionReasoningScore

Conciseness

The body is dense with rules and provenance and never explains basics Claude already knows, but the three-paragraph motivational intro (port vs. greenfield oracle framing), the "Field origin" paragraph, and rhetorical asides ("it is a slot machine", "the door the next flake walks through") are tokens that could be trimmed — matching anchor 4's minor over-explanation rather than anchor 5's every-token-earns-its-place.

4 / 5

Actionability

Guidance is concrete and executable at the instruction level — "Hash through explicit `to_le_bytes()`", "store the bit pattern: `enum Literal { F64Bits(u64), .. }`", "`RUSTFLAGS=\"--cfg loom\"`", "delta-based assertions — snapshot the counter before, run, assert the difference" — but no complete code skeleton (e.g., of the single entropy fixture) is provided, leaving minor gaps that keep it below anchor 5.

4 / 5

Workflow Clarity

The sections form a coherent diagnostic progression (entropy funnel → silent hashing sources → golden mechanics → flaky causes → seed logging → cross-arch → concurrency) with explicit checkpoints embedded ("Checksum-gate before you enumerate", "A golden change is a proof obligation... Record an isomorphism note", "Throw them away, then measure"), and the quasi-destructive re-bless path does carry validation, so no cap applies — though it is a rules catalogue rather than a stepwise workflow with error-recovery loops.

4 / 5

Progressive Disclosure

A single self-contained file (~170 lines) with clearly numbered, well-cross-linked sections (§1–§7 referenced from later sections and the smells list); no bundle files exist, and references to sibling skills (testing-strategy, golden-parity, verification-ratchet) are named inline. At this length a couple of sections (e.g., §6 cross-architecture detail) could arguably split out, which keeps it at anchor 4 rather than 5.

4 / 5

Total

16

/

20

Passed

Description

82%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong description with an explicit, multi-trigger "Use when" clause, a concrete coverage list, and explicit boundary-setting against a sibling skill. Its main deductions are the second-person "your own goldens" aside and a coverage list of topic nouns rather than action verbs.

Suggestions

Rewrite in consistent third person (e.g., "the project's own goldens are the spec") to avoid the voice penalty and save description tokens.

Lead the coverage list with action verbs (e.g., "Funnels all entropy through one fixture, replaces HashMap iteration with BTreeMap ordering, checksum-gates golden comparisons") rather than topic nouns.

Trim the motivational middle sentence; the trigger and coverage clauses already carry the information.

DimensionReasoningScore

Specificity

The description lists six concrete covered topics ("single-entropy-fixture funnel, stable hashing/ordering, golden checksum-gating, the four flaky-test causes, seeded randomness, and cross-architecture drift"), which is anchor-4 territory, but they are framed as topics covered rather than actions, and the second-person phrasing "your own goldens ARE the spec" triggers the rubric's 1-point voice penalty, bringing it to 3.

3 / 5

Completeness

Both questions are explicitly answered: the "what" via "Covers the single-entropy-fixture funnel, stable hashing/ordering, golden checksum-gating..." and the "when" via "Use when writing deterministic tests, golden/snapshot suites, or any Rust whose output must be byte-stable across runs, machines, and parallel test execution — or when a test is flaky" — concrete trigger phrases matching the anchor-5 example.

5 / 5

Trigger Term Quality

Natural user phrases are present — "deterministic tests", "golden/snapshot suites", "when a test is flaky", "byte-stable", "parallel test execution" — but a few common variants are missing and terms like "greenfield" and "reference oracle" are author jargon rather than user-said language, so it sits between anchors 4 and 5, below comprehensive.

4 / 5

Distinctiveness Conflict Risk

It carves a clear Rust-determinism niche and explicitly disambiguates from the closest sibling ("testing-strategy says WHERE determinism is injected; this is the Rust-deep HOW"), so conflict risk with other testing skills is minimal.

5 / 5

Total

17

/

20

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository

igouss/gitweb-rs

Path

.claude/skills/rust-determinism-engineering/SKILL.md

Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.