
This is a meta-router that decides which specialized Rust skill Claude should load based on your question. It works in three layers: language mechanics (ownership, lifetimes, error codes), design patterns (performance, anti-patterns, domain modeling), and domain constraints (fintech, web, embedded). When you hit a borrow checker error or ask about async patterns, it traces through the layers to pull in the right combination of skills. The negotiation protocol is interesting: it forces Claude to show confidence levels and gaps when you ask comparative questions like "tokio vs async-std" or request best practices. It's essentially a dispatch table with 30+ routing rules, error code lookups, and domain keyword detection. If you work in Rust and want more structured responses that acknowledge what they don't cover, this shapes how Claude approaches your questions.
$ npx -y skills add actionbook/rust-skills --skill rust-router --agent claude-codeInstalls into .claude/skills of the current project.
Version: 2.0.0 | Last Updated: 2025-01-22
v2.0: Context optimized - detailed examples moved to sub-files
Don't answer directly. Trace through the cognitive layers first.
Layer 3: Domain Constraints (WHY)
├── Business rules, regulatory requirements
├── domain-fintech, domain-web, domain-cli, etc.
└── "Why is it designed this way?"
Layer 2: Design Choices (WHAT)
├── Architecture patterns, DDD concepts
├── m09-m15 skills
└── "What pattern should I use?"
Layer 1: Language Mechanics (HOW)
├── Ownership, borrowing, lifetimes, traits
├── m01-m07 skills
└── "How do I implement this in Rust?"
| User Signal | Entry Layer | Direction | First Skill |
|---|---|---|---|
| E0xxx error | Layer 1 | Trace UP ↑ | m01-m07 |
| Compile error | Layer 1 | Trace UP ↑ | Error table below |
| "How to design..." | Layer 2 | Check L3, then DOWN ↓ | m09-domain |
| "Building [domain] app" | Layer 3 | Trace DOWN ↓ | domain-* |
| "Best practice..." | Layer 2 | Both directions | m09-m15 |
| Performance issue | Layer 1 → 2 | UP then DOWN | m10-performance |
When domain keywords are present, you MUST load BOTH skills:
| Domain Keywords | L1 Skill | L3 Skill |
|---|---|---|
| Web API, HTTP, axum, handler | m07-concurrency | domain-web |
| 交易, 支付, trading, payment | m01-ownership | domain-fintech |
| CLI, terminal, clap | m07-concurrency | domain-cli |
| kubernetes, grpc, microservice | m07-concurrency | domain-cloud-native |
| embedded, no_std, MCU | m02-resource | domain-embedded |
BEFORE answering, check if negotiation is required:
| Query Contains | Action |
|---|---|
| "比较", "对比", "compare", "vs", "versus" | MUST use negotiation |
| "最佳实践", "best practice" | MUST use negotiation |
| Domain + error (e.g., "交易系统 E0382") | MUST use negotiation |
| Ambiguous scope (e.g., "tokio 性能") | SHOULD use negotiation |
When negotiation is required, include:
## Negotiation Analysis
**Query Type:** [Comparative | Cross-domain | Synthesis | Ambiguous]
**Negotiation:** Enabled
### Source: [Agent/Skill Name]
**Confidence:** HIGH | MEDIUM | LOW | UNCERTAIN
**Gaps:** [What's missing]
## Synthesized Answer
[Answer]
**Overall Confidence:** [Level]
**Disclosed Gaps:** [Gaps user should know]
详细协议见:
patterns/negotiation.md
When creating new Rust projects or Cargo.toml files, ALWAYS use:
[package]
edition = "2024" # ALWAYS use latest stable edition
rust-version = "1.85"
[lints.rust]
unsafe_code = "warn"
[lints.clippy]
all = "warn"
pedantic = "warn"
| Pattern | Route To |
|---|---|
| move, borrow, lifetime, E0382, E0597 | m01-ownership |
| Box, Rc, Arc, RefCell, Cell | m02-resource |
| mut, interior mutability, E0499, E0502, E0596 | m03-mutability |
| generic, trait, inline, monomorphization | m04-zero-cost |
| type state, phantom, newtype | m05-type-driven |
| Result, Error, panic, ?, anyhow, thiserror | m06-error-handling |
| Send, Sync, thread, async, channel | m07-concurrency |
| unsafe, FFI, extern, raw pointer, transmute | unsafe-checker |
| Pattern | Route To |
|---|---|
| domain model, business logic | m09-domain |
| performance, optimization, benchmark | m10-performance |
| integration, interop, bindings | m11-ecosystem |
| resource lifecycle, RAII, Drop | m12-lifecycle |
| domain error, recovery strategy | m13-domain-error |
| mental model, how to think | m14-mental-model |
| anti-pattern, common mistake, pitfall | m15-anti-pattern |
| Domain Keywords | Route To |
|---|---|
| fintech, trading, decimal, currency | domain-fintech |
| ml, tensor, model, inference | domain-ml |
| kubernetes, docker, grpc, microservice | domain-cloud-native |
| embedded, sensor, mqtt, iot | domain-iot |
| web server, HTTP, REST, axum, actix | domain-web |
| CLI, command line, clap, terminal | domain-cli |
| no_std, microcontroller, firmware | domain-embedded |
| Error Code | Route To | Common Cause |
|---|---|---|
| E0382 | m01-ownership | Use of moved value |
| E0597 | m01-ownership | Lifetime too short |
| E0506 | m01-ownership | Cannot assign to borrowed |
| E0507 | m01-ownership | Cannot move out of borrowed |
| E0515 | m01-ownership | Return local reference |
| E0716 | m01-ownership | Temporary value dropped |
| E0106 | m01-ownership | Missing lifetime specifier |
| E0596 | m03-mutability | Cannot borrow as mutable |
| E0499 | m03-mutability | Multiple mutable borrows |
| E0502 | m03-mutability | Borrow conflict |
| E0277 | m04/m07 | Trait bound not satisfied |
| E0308 | m04-zero-cost | Type mismatch |
| E0599 | m04-zero-cost | No method found |
| E0038 | m04-zero-cost | Trait not object-safe |
| E0433 | m11-ecosystem | Cannot find crate/module |
| Pattern | Route To | Action |
|---|---|---|
| latest version, what's new | rust-learner | Use agents |
| API, docs, documentation | docs-researcher | Use agent |
| code style, naming, clippy | coding-guidelines | Read skill |
| unsafe code, FFI | unsafe-checker | Read skill |
| code review | os-checker | See integrations/os-checker.md |
| Keyword | Resolution |
|---|---|
unsafe | unsafe-checker (more specific than m11) |
error | m06 for general, m13 for domain-specific |
RAII | m12 for design, m01 for implementation |
crate | rust-learner for version, m11 for integration |
tokio | tokio-* for API, m07 for concepts |
Priority Hierarchy:
1. Error codes (E0xxx) → Direct lookup, highest priority
2. Negotiation triggers (compare, vs, best practice) → Enable negotiation
3. Domain keywords + error → Load BOTH domain + error skills
4. Specific crate keywords → Route to crate-specific skill if exists
5. General concept keywords → Route to meta-question skill
| File | Content |
|---|---|
patterns/negotiation.md | Negotiation protocol details |
examples/workflow.md | Workflow examples |
integrations/os-checker.md | OS-Checker integration |
Sponsors