
The m06-error-handling skill teaches Rust developers how to choose appropriate error handling strategies by distinguishing between expected failures, normal absence, and bugs—using Result, Option, panic, and crates like thiserror and anyhow. It's designed for developers writing both library and application code who need to decide when to propagate errors, return Options, panic, or add context to failures. The skill solves the problem of unclear error handling decisions by providing a decision framework that traces back to domain design and down to specific implementation patterns.
$ npx -y skills add zhanghandong/rust-skills --skill m06-error-handling --agent claude-codeInstalls into .claude/skills of the current project.
Layer 1: Language Mechanics
Is this failure expected or a bug?
Before choosing error handling strategy:
| Pattern | Don't Just Say | Ask Instead |
|---|---|---|
| unwrap panics | "Use ?" | Is None/Err actually possible here? |
| Type mismatch on ? | "Use anyhow" | Are error types designed correctly? |
| Lost error context | "Add .context()" | What does the caller need to know? |
| Too many error variants | "Use Box" | Is error granularity right? |
Before handling an error:
What kind of failure is this?
Who handles this?
What context is needed?
When error strategy is unclear:
"Should I return Result or Option?"
↑ Ask: Is absence/failure normal or exceptional?
↑ Check: m09-domain (what does domain say?)
↑ Check: domain-* (error handling requirements)
| Situation | Trace To | Question |
|---|---|---|
| Too many unwraps | m09-domain | Is the data model right? |
| Error context design | m13-domain-error | What recovery is needed? |
| Library vs app errors | m11-ecosystem | Who are the consumers? |
From design to implementation:
"Expected failure, library code"
↓ Use: thiserror for typed errors
"Expected failure, application code"
↓ Use: anyhow for ergonomic errors
"Absence is normal (find, get, lookup)"
↓ Use: Option<T>
"Bug or invariant violation"
↓ Use: panic!, assert!, unreachable!
"Need to propagate with context"
↓ Use: .context("what was happening")
| Pattern | When | Example |
|---|---|---|
Result<T, E> | Recoverable error | fn read() -> Result<String, io::Error> |
Option<T> | Absence is normal | fn find() -> Option<&Item> |
? | Propagate error | let data = file.read()?; |
unwrap() | Dev/test only | config.get("key").unwrap() |
expect() | Invariant holds | env.get("HOME").expect("HOME set") |
panic! | Unrecoverable | panic!("critical failure") |
| Context | Error Crate | Why |
|---|---|---|
| Library | thiserror | Typed errors for consumers |
| Application | anyhow | Ergonomic error handling |
| Mixed | Both | thiserror at boundaries, anyhow internally |
Is failure expected?
├─ Yes → Is absence the only "failure"?
│ ├─ Yes → Option<T>
│ └─ No → Result<T, E>
│ ├─ Library → thiserror
│ └─ Application → anyhow
└─ No → Is it a bug?
├─ Yes → panic!, assert!
└─ No → Consider if really unrecoverable
Use ? → Need context?
├─ Yes → .context("message")
└─ No → Plain ?
| Error | Cause | Fix |
|---|---|---|
unwrap() panic | Unhandled None/Err | Use ? or match |
| Type mismatch | Different error types | Use anyhow or From |
| Lost context | ? without context | Add .context() |
cannot use ? | Missing Result return | Return Result<(), E> |
| Anti-Pattern | Why Bad | Better |
|---|---|---|
.unwrap() everywhere | Panics in production | .expect("reason") or ? |
| Ignore errors silently | Bugs hidden | Handle or propagate |
panic! for expected errors | Bad UX, no recovery | Result |
| Box everywhere | Lost type info | thiserror |
| When | See |
|---|---|
| Domain error strategy | m13-domain-error |
| Crate boundaries | m11-ecosystem |
| Type-safe errors | m05-type-driven |
| Mental models | m14-mental-model |
Sponsors