rust-stinger skill
scriptsCargo manifest, async service, CLI/TUI, persistence layer을 포함한 Rust 코드 구현, 리팩터링 및 리뷰를 수행합니다.
- GitHub 스타
- 95
- 파일
- 32
- 마지막 커밋
- 2026년 10월 18일
- 라이선스
- AGPL-3.0-or-later
- 허용된 도구
- —
rust-stinger skill은 무엇을 하나요?
Cargo manifest, async service, CLI/TUI, persistence layer을 포함한 Rust 코드 구현, 리팩터링 및 리뷰를 수행합니다.
skill의 SKILL.md에서 생성된 요약입니다.
rust-stinger skill 정보
Rust 소스 코드를 개발, 리팩터링 및 리뷰하는 것이 이 SKILL의 주요 기능입니다. Cargo manifest, async service, CLI 및 TUI를 작업하는 개발자를 지원합니다. 이 도구는 workspace, build script, testing suite를 포함한 복잡한 Rust 생태계를 관리하는 데 특화되어 있으며, PRD 및 ADR과 같은 프로젝트 요구 사항을 엄격히 준수합니다.
에이전트가 이 SKILL을 로드하면 Tokio, Axum, SQLx, Ratatui와 같은 라이브러리를 사용하여 컴포넌트를 설계하고 구현할 수 있습니다. async stream, SQLite를 사용한 persistence layer 및 상태 전환의 정확성을 검증하는 데 집중합니다. 결정 사항이 안전성, 자격 증명 또는 비용 발생에 영향을 미치는 경우, 구현이 정의된 범위를 벗어나지 않도록 작업을 중단하고 명확한 설명을 요청하도록 설계되었습니다.
일반적인 코딩 어시스턴트와 달리, 이 SKILL은 기계적인 리뷰 가능성과 릴리스 증거 생성에 중점을 둡니다. 컴파일러 및 Cargo 업그레이드, edition 마이그레이션, MSRV 정책 업데이트를 처리합니다. unsafe code를 사용하지 않는 것을 기본 원칙으로 하며, 파괴적이거나 외부적인 효과를 발생시키는 모든 작업에 대해 명시적인 권한을 요구하는 엄격한 제약 조건 하에 작동합니다.
agent가 수행할 수 있는 작업
- 1async service 리팩터링task cancellation, timeout, backpressure를 포함한 async stream의 동작을 구현하고 검증합니다.
- 2Rust 툴체인 관리컴파일러 또는 Cargo 업그레이드를 수행하고 edition 또는 resolver 마이그레이션을 관리합니다.
- 3persistence layer 구현SQLx를 사용하여 SQLite 트랜잭션, 마이그레이션 및 crash recovery를 개발하고 검증합니다.
- 4CLI 및 TUI 도구 구축스크립트 작성이 가능한 CLI 및 TUI 인터페이스를 생성합니다.
skill의 SKILL.md를 기반으로 작성되었습니다. 모든 사용 사례는 해당 문서의 인용구와 대조 확인되었습니다.
rust-stinger skill 설치 방법
각 명령은 skill 폴더를 클라이언트의 skills 디렉터리에 복사합니다. 클라이언트를 재시작하거나 새 세션을 시작하여 로드하세요.플러그인에 rust-stinger skill 설치
/plugin marketplace add legioncodeinc/vibe-coding-tools /plugin install backend-and-desktop-tools@wasp-nest
Claude Code에 rust-stinger skill 설치
~/.claude/skills/rust-stinger/SKILL.md
$ npx degit legioncodeinc/vibe-coding-tools/plugins/backend-and-desktop-tools/skills/rust-stinger ~/.claude/skills/rust-stingerClaude Code (project)에 rust-stinger skill 설치
.claude/skills/rust-stinger/SKILL.md
$ npx degit legioncodeinc/vibe-coding-tools/plugins/backend-and-desktop-tools/skills/rust-stinger .claude/skills/rust-stingerCodex에 rust-stinger skill 설치
~/.agents/skills/rust-stinger/SKILL.md
$ npx degit legioncodeinc/vibe-coding-tools/plugins/backend-and-desktop-tools/skills/rust-stinger ~/.agents/skills/rust-stingerGemini CLI에 rust-stinger skill 설치
~/.gemini/skills/rust-stinger/SKILL.md
$ npx degit legioncodeinc/vibe-coding-tools/plugins/backend-and-desktop-tools/skills/rust-stinger ~/.gemini/skills/rust-stingerClaude.ai에 rust-stinger skill 설치
npx degit legioncodeinc/vibe-coding-tools/plugins/backend-and-desktop-tools/skills/rust-stinger rust-stinger && zip -r rust-stinger.zip rust-stinger # claude.ai → Customize → Skills → + → Create skill → Upload a skill → rust-stinger.zip
자동 검사
2026년 11월 14일에 실행됨전체 보고서 →아직 검증되지 않았습니다. 스킬은 점수 80, 심각한 발견 없음, 검사에 충분한 데이터가 필요합니다. 검사 방식 · 점수를 높이기 위해 수정할 항목
- 소스 저장소 연결 및 접근 가능legioncodeinc/vibe-coding-tools/plugins/backend-and-desktop-tools/skills/rust-stinger is public and reachablegithub.com/legioncodeinc/vibe-coding-tools/tree/HEAD/plugins ↗통과
- 오픈 소스 라이선스 (OSI)AGPL-3.0-or-later is not OSI-approvedspdx.org/licenses/AGPL-3.0-or-later.html ↗결함 발견
- 최근 1년 내 커밋 또는 릴리스Last commit 2026-10-18github.com/legioncodeinc/vibe-coding-tools/commits ↗통과
- 스크립트에 다운로드 후 실행이나 자격 증명 탈취가 없음1 script read; no download-and-run, remote eval, base64 execution or credential upload통과
- SKILL.md에 숨겨진 지침 없음No override phrases, safeguard switches, credential requests or hidden charactersgithub.com/legioncodeinc/vibe-coding-tools/blob/HEAD/plugins ↗통과
- 사전 허용 도구가 제한됨No pre-approved tools; the agent asks before using onegithub.com/legioncodeinc/vibe-coding-tools/blob/HEAD/plugins ↗통과
- SKILL.md가 Agent Skills 형식을 따름Valid name and description; says when to use it; 146 linesgithub.com/legioncodeinc/vibe-coding-tools/blob/HEAD/plugins ↗통과
SKILL.md
GitHub에서 보기Rust Stinger
Start with README.md for the workflow map and detailed references.
Purpose
Equip rust-wasp-drone to own bounded Rust implementation while preserving the authority of the exact PRD, ADR, ledger, and peer specialists. Make async ownership, durability, state transitions, redaction, and release evidence mechanically reviewable. Stop at a recorded fail-closed boundary whenever a missing decision affects safety, compatibility, money, credentials, signing, publication, or another external effect.
When to use
- Implement, refactor, debug, or review Rust source, Cargo manifests, workspaces, features, build scripts, or tests.
- Design or verify Tokio, Axum, Tower, SQLx, SQLite, Clap, Ratatui, tracing, rustls, and Rust release-evidence paths.
- Ask for the newest stable Rust information, a compiler/Cargo upgrade, an edition or resolver migration, or an MSRV policy.
- Prepare an AI-assisted contribution to
rust-lang/rustwhen the owning team has ratified the upstream project's live LLM policy.
When not to use
- Tauri application-shell, capability, plugin, updater, and webview integration work belongs to
tauri-stinger, with this Stinger owning the Rust implementation underneath. - Dependency advisory and license disposition belongs to
dependency-audit-stinger; this Stinger generates Rust-specific evidence but does not accept risk. - Database schema architecture belongs to
db-stinger; this Stinger implements and proves an approved SQLx or SQLite design. - General AI model/provider selection belongs to
ai-tools-platform-stinger, and cognitive/RAG architecture belongs tomind-stinger.
Required inputs
Before editing, obtain:
- the exact repository or worktree, owned paths, and concurrent-work boundaries;
- the authorizing PRD/ADR/ledger rows, acceptance criteria, gates, and repository instructions;
- the existing Cargo graph, toolchain files, feature/target matrix, migrations, code, tests, and release configuration;
- approved protocol, provider, persistence, security, CLI, platform, and operational contracts;
- the required verification commands and explicit authorization for any external effect.
If a missing input determines safety, public compatibility, monetary behavior, credentials, publication, or destructive behavior, return a blocker rather than choosing silently.
Procedure
- Reconstruct authority, scope, gate state, and worktree ownership using
guides/00-authority-and-principles.md. - Inventory the Rust system and record revalidation points using
guides/01-inspect-workspace.md. - If the request depends on current or upcoming Rust behavior, refresh the version claim using
guides/10-refresh-current-rust.mdbefore selecting a design. - Establish the smallest coherent crate, feature, type, and error design using
guides/02-design-workspace-and-types.md. - Implement a test-first acceptance slice using
guides/03-implement-bounded-slices.md. - Prove task, cancellation, stream, backpressure, timeout, retry, and shutdown behavior using
guides/04-prove-async-streams.md. For native CPU, blocking I/O or persistent background workers, first classify the execution boundary withguides/11-native-background-work.mdandtemplates/native-background-worksheet.md. - Prove SQLite/SQLx transactions, migrations, crash recovery, and typed state transitions using
guides/05-prove-persistence-and-state.md. - Implement provider and harness edges only behind approved contracts using
guides/06-implement-adapters.md. - Build scriptable CLI and optional TUI surfaces using
guides/07-build-cli-and-tui.md. - Run the complete owned verification and generate release evidence without publishing using
guides/08-verify-and-package-evidence.md. - Produce the acceptance-linked handoff, route Security before Quality, and leave unresolved gates explicit using
guides/09-close-the-loop.md.
Operating constraints
- Treat the named PRD, ADR, ledger, gate state, and repository instructions as authority. Never start blocked or deferred work.
- Keep provider code subordinate to the host harness: it may route inference but never acquire tools, approvals, repository access, memory, or user interaction.
- Default to bounded queues, explicit task owners, operation-level cancellation review, and no transparent replay after visible output or a harness-visible tool call.
- Put monetary/quota checks, reservations, reconciliation, idempotency, and aggregate updates in explicit transactions with concurrency and crash evidence.
- Keep credentials in approved secret references and start sensitive instrumentation with
skip_all; allowlist only non-sensitive correlation and state fields. - Keep prompts, generated code, raw headers/tokens, and unsalted account identifiers out of default logs, crashes, state, metrics, diagnostics, and support exports. Enforce approved egress allowlists and never bypass redirect, DNS, or SSRF checks.
- Default to no
unsafe. Any exception needs a minimal scope, written invariant, targeted tests, and independent review. - Use fake providers and fixtures by default. Do not use live credentials, paid traffic, subscriptions, signing identities, publishing, auto-update, Git initialization, or destructive controls without explicit authorization.
- Never promote retrieval-time crate/tool versions, OS targets, durability settings, timeout values, signing systems, or soak thresholds into timeless defaults. Revalidate them at the decision point.
- Preserve peer authority: protocol semantics, product/provider policy, security acceptance, schema review, dependency/license disposition, CI topology, signing/publication, and final PRD Quality remain separate handoffs.
- Preserve verification order: implementation checks, Security, affected-check reruns, then Quality. Partial or stale evidence is not shipped evidence.
See guides/00-authority-and-principles.md for the rationale and fail-closed decision table.
Output contract
Produce a Rust implementation handoff from templates/implementation-handoff.md, backed by exact commands and artifacts. When release work is in scope, also populate templates/release-evidence-manifest.yaml without signing or publishing unless independently authorized. Store actual execution and gate reports in the active repository's root library/ hierarchy. The Stinger's reports/ folder contains reusable report-shape templates only.
Examples
-
examples/01-happy-path-bounded-service-slice.md- Cargo/Axum/SQLx acceptance slice with deterministic evidence. -
examples/02-edge-visible-output-cancellation.md- cancellation after visible output, where replay must remain disabled. -
examples/03-edge-concurrent-budget-reservation.md- concurrent SQLite reservation and crash boundary. -
examples/04-release-evidence-with-closed-gates.md- package evidence generation with signing/publication held closed. -
examples/05-rust-1-98-refresh.md- bounded upgrade evidence from 1.97.1 to the fixed 1.98.1 patch. -
examples/06-native-background-worker.md- bounded native worker ownership with an offline standard-library fixture, typed errors, cancellation and joined shutdown.
Research and refresh points
For current toolchain, Cargo, security, nightly, or upstream AI-policy claims, read references/research/distilled-rust-current.md first. Its raw archive separates bounded primary-source captures from Stinger interpretation and cites every current source record. For the broader architecture corpus, read research/research-summary.md and research/evidence-synthesis.md; the index and legacy primary notes behind them are source-repository research, not shipped with the plugin. Re-fetch the Rust release/platform matrix, Tokio support policy, the selected SQLx transaction API, stable rustls APIs, cargo-dist behavior, and RustSec data at the point of use. Both research dates are snapshots, not permanent defaults.
Human decisions that remain open
Do not resolve these from the Stinger alone:
- supported OS/architecture/install matrix and minimum OS baselines;
- actual MSRV after the resolved graph and public features exist;
- SQLite power-loss durability and contended-writer behavior;
- durable replay/promotion event schema;
- first-milestone TUI scope;
- signing identity, attestation platform, installer formats, and publication authorization;
- quantitative soak pass/fail thresholds.
Record the missing owner/decision, affected acceptance criteria, and first authorized next action in the handoff.
Paired owner
This Stinger is paired with rust-wasp-drone.
References map
TOPIC.md- load when auditing the forge scope, ownership boundary, or required outcomes.references/REFERENCE.md- load first when the task needs the deep reference layer rather than the root procedure alone.references/CURRENT-RUST.md- load for current stable, compiler/Cargo upgrades, edition/resolver changes, MSRV decisions, and security-driven toolchain floors.references/NIGHTLY-WATCHLIST.md- load when the user asks about upcoming Rust features; never use it as stable production guidance.references/UPSTREAM-RUST-LLM-POLICY.md- load only for AI-assisted contributions torust-lang/rustin a team covered by the upstream policy.references/research/distilled-rust-current.md- load when a current domain claim needs verification or sources disagree.- Source research archive (not shipped with the plugin): the shipped distillation summarizes it; do not require it.
research/research-summary.mdandresearch/evidence-synthesis.md- load for the broader 2026-07-24 architecture and tooling corpus retained from the original pair. The index and the per-topic primary notes are source-repository research, not shipped with the plugin.scripts/inspect-rust-workspace.py- run for a deterministic static Cargo/version/unsafe inventory before planning a workspace change.guides/10-refresh-current-rust.md- load for any request using "latest", "current", "upgrade Rust", "MSRV", or nightly language.
Related drones and stingers
- tauri-stinger - Tauri 2 app-shell, IPC, capability, sidecar, plugin, updater, and distribution work built on Rust.
- dependency-audit-stinger - dependency advisories, lockfile hygiene, license decisions, SBOMs, and supply-chain risk.
- security-stinger - independent vulnerability review and security acceptance.
- db-stinger - database schema, migration, and indexing authority.
- rust-wasp-drone - the paired implementation and review agent for this Stinger.
Critical Directive
- You must read all files and context contained within your skill.
- In the event your core knowledge does not provide sufficient guidance you must make every attempt to search the internet, related knowledge base documentation files, and other available resources to supplement your knowledge prior to proceeding with your task.
- Additional related skills can be found here:
- tauri-stinger - Tauri 2 application integration and AI desktop/mobile shell patterns.
- dependency-audit-stinger - Rust dependency, advisory, license, and supply-chain decisions.
- security-stinger - Independent security audit and remediation.
Ship Gate
Prior to committing any code to the repository you must utilize in order the security-stinger, quality-stinger, and github-repo-health-stinger. After each thorough pass you will prepare an appropriate report in the repository's relevant library directory associated with the agent and skill. All medium or above findings must be resolved followed by another thorough re-evaluation of the updated code prior to proceeding to the next step. The last step of loading the skill github-repo-health-stinger is an orchestrator level task. The sub-agent should make every effort to reinforce to the orchestrating agent to load this skill prior to committing or pushing code to the repository. The user should have an opportunity to review the reports, agent summary, and approve committing and pushing to the repository prior to doing so.
이 스킬의 파일
32rust-stinger skill 질문
Claude Code에 rust-stinger skill을 어떻게 설치하나요?
npx degit legioncodeinc/vibe-coding-tools/plugins/backend-and-desktop-tools/skills/rust-stinger ~/.claude/skills/rust-stinger를 실행합니다. Claude Code는 ~/.claude/skills(모든 프로젝트) 또는 .claude/skills(한 프로젝트)에서 skills를 로드하며, 설명이 작업과 일치할 때 skill을 사용합니다.
rust-stinger skill을 사용할 수 있는 클라이언트는 무엇인가요?
Agent Skills 형식을 읽는 모든 클라이언트(Claude Code, Codex, Gemini CLI, claude.ai(ZIP으로 업로드))가 사용할 수 있습니다. skill 내 스크립트는 클라이언트가 코드 실행을 허용하는 경우에만 실행됩니다.
어떤 특정 Rust 기술을 다룰 수 있나요?
Tokio, Axum, Tower, SQLx, SQLite, Clap, Ratatui, tracing, rustls를 지원하도록 설계되었습니다.
Tauri 애플리케이션 셸 통합을 처리할 수 있나요?
아니요, Tauri application-shell, capability, plugin, updater 및 webview 통합 작업은 tauri-stinger에서 담당합니다.
데이터베이스 스키마 아키텍처는 어떻게 처리하나요?
스키마를 직접 설계하지는 않으며, db-stinger가 승인한 설계를 구현하고 검증합니다.