npx skills add smithery/yanko-belov --skill fail-fast
yanko-belov/code-craft
fail-fast
Use when handling errors. Use when tempted to catch and swallow exceptions. Use when returning default values to hide failures.
Installation
npx skills add yanko-belov/code-craft --skill fail-fast
Similar popular skills
Related neighbors and high-traction skills in the same topics — useful to compare before installing.
Troubleshoots failing applications by discovering and analyzing CloudWatch log groups to identi…
4.8K installsRuns and interprets AWS Resilience Hub v2 failure mode assessments. Covers starting assessments…
2.3K installsDebug a Pulumi update or preview that failed: read the failure Pulumi already recorded, find wh…
1 installsRead a liarjs fingerprint report and attribute each failing check to the component that produce…
43.4K installsDiagnose CI failures for a PR using the GitHub CLI, extract error logs, and generate a plan to …
24.4K installsUse when writing or touching any error path, catch block, fallback, default value, retry, or de…
5.3K installsAlso in this package
Other skills from yanko-belov/code-craft · top by installs.
npx skills add yanko-belov/code-craft
More details
Agent compatibility
Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.
Also listed on
Alternate registries and mirrors of this skill.
Repository health
master
Package contents
Files included with this skill beyond the listing page.
-
skill md
SKILL.md7,563 B -
docs
SUMMARY.md144 B
History
- First seen on skills.sh
- First recorded snapshot · 47 installs
SKILL.md
Fail Fast
Overview
When something goes wrong, fail immediately and visibly.
Don't hide errors with try/catch that returns defaults. Don't let invalid state propagate. Fail at the point of failure, not three layers later with corrupted data.
When to Use
- Writing error handling code
- Tempted to catch and return default
- Adding "defensive" null checks everywhere
- Wrapping everything in try/catch
- Returning error objects instead of throwing
The Iron Rule
NEVER hide failures. Fail loud, fail early.
No exceptions:
- Not for "the app shouldn't crash"
- Not for "return something rather than throw"
- Not for "handle errors gracefully"
- Not for "defensive programming"
Detection: The "Swallow" Smell
If errors disappear silently, you're failing slow:
// ❌ VIOLATION: Hiding failures
async function processPayment(userId: string, amount: number): Promise<PaymentResult> {
try {
const user = await getUser(userId);
if (!user) return { success: false, error: 'User not found' };
const card = await validateCard(user.cardToken);
if (!card.valid) return { success: false, error: 'Invalid card' };
const result = await chargeCard(card, amount);
if (!result.success) return { success: false, error: 'Payment failed' };
return { success: true, transactionId: result.id };
} catch (error) {
return { success: false, error: 'Internal error' }; // ← SWALLOWED!
}
}
Problems:
- Caller doesn't know WHAT failed
- Stack trace is lost
- Bugs hide as "internal error"
- No visibility into actual failures
The Correct Pattern: Fail Fast
Throw at the point of failure. Let errors propagate:
// ✅ CORRECT: Fail fast
async function processPayment(userId: string, amount: number): Promise<Transaction> {
// Validate early - fail fast on bad input
if (!userId) throw new ValidationError('userId is required');
if (amount <= 0) throw new ValidationError('amount must be positive');
// Let failures propagate - don't swallow
const user = await getUser(userId);
if (!user) throw new NotFoundError(`User ${userId} not found`);
const card = await validateCard(user.cardToken);
if (!card.valid) throw new PaymentError('Card validation failed', card.errors);
// This might throw - that's okay! Let it.
const transaction = await chargeCard(card, amount);
return transaction;
}
// Caller handles errors appropriately
try {
const tx = await processPayment(userId, amount);
res.json({ success: true, transactionId: tx.id });
} catch (error) {
if (error instanceof ValidationError) {
res.status(400).json({ error: error.message });
} else if (error instanceof NotFoundError) {
res.status(404).json({ error: error.message });
} else if (error instanceof PaymentError) {
res.status(402).json({ error: error.message });
} else {
// Unknown error - log it, return 500
logger.error('Payment failed', error);
res.status(500).json({ error: 'Internal server error' });
}
}
Why Fail-Slow Is Dangerous
| Problem | Impact |
|---|---|
| Hidden bugs | Errors become "it didn't work" |
| Lost context | Stack trace shows catch, not cause |
| Corrupted state | Invalid data propagates |
| Debugging nightmare | Where did it actually fail? |
| Silent data loss | Operations fail but app continues |
Fail Fast Techniques
1. Validate Early
function createUser(data: unknown): User {
// Fail IMMEDIATELY on bad input
if (!data || typeof data !== 'object') {
throw new ValidationError('Invalid user data');
}
const { email, name } = data as Record<string, unknown>;
if (!email || typeof email !== 'string') {
throw new ValidationError('Email is required');
}
if (!name || typeof name !== 'string') {
throw new ValidationError('Name is required');
}
// Only proceed with valid data
return new User(email, name);
}
2. Assert Invariants
function withdraw(account: Account, amount: number): void {
// Assert what must be true
assert(amount > 0, 'Withdrawal amount must be positive');
assert(account.balance >= amount, 'Insufficient funds');
account.balance -= amount;
// Post-condition check
assert(account.balance >= 0, 'Balance went negative - invariant violated');
}
3. Use Type System
// ❌ Fail slow: null checks everywhere
function processOrder(order: Order | null): void {
if (!order) return; // Silent failure
// ...
}
// ✅ Fail fast: require valid input
function processOrder(order: Order): void {
// If order is null, TypeScript catches it
// If it gets here with null, it will throw - good!
}
Pressure Resistance Protocol
1. "The App Shouldn't Crash"
Pressure: "Users will see errors if we throw"
Response: Users seeing a clear error is better than corrupted data or silent failure.
Action: Throw errors, catch at boundaries (API layer), return appropriate HTTP codes.
2. "Return Something Rather Than Throw"
Pressure: "Returning error objects is more functional"
Response: Error objects are fine IF callers check them. They usually don't.
Action: Throw for unexpected failures. Use Result types only if callers actually handle both cases.
3. "Handle Errors Gracefully"
Pressure: "Graceful = don't throw"
Response: Graceful = appropriate response. Swallowing is not graceful.
Action: Throw, catch at boundary, return meaningful error response.
4. "Defensive Programming"
Pressure: "Defensive code handles all cases"
Response: Defensive = validate early and fail. Not = hide failures.
Action: Validate inputs, assert invariants, throw on violations.
Red Flags - STOP and Reconsider
If you notice ANY of these, refactor:
catch (e) { return null; }catch (e) { return { success: false }; }if (!x) return;(silent early return)try { } catch { }(empty catch)- Returning default values on error
- "Error: Internal error" (generic catch-all)
- Logs error but continues execution
All of these mean: Let the error propagate or throw explicitly.
Quick Reference
| Fail Slow (Bad) | Fail Fast (Good) |
|---|---|
catch (e) { return null } |
catch (e) { throw e } |
if (!user) return |
if (!user) throw new NotFoundError() |
return { success: false } |
throw new OperationError() |
| Generic "internal error" | Specific error types |
| Swallow and continue | Propagate and handle at boundary |
Common Rationalizations (All Invalid)
| Excuse | Reality |
|---|---|
| "App shouldn't crash" | Clear errors are better than hidden bugs. |
| "Return instead of throw" | Callers ignore return values. Throws can't be ignored. |
| "Graceful error handling" | Swallowing isn't graceful. |
| "Defensive programming" | Defensive = validate and fail, not hide. |
| "Never let functions crash" | Crashing on errors finds bugs. |
| "User experience" | Users prefer "payment failed" over silent failures. |
The Bottom Line
Fail fast. Fail loud. Fail at the source.
When errors occur: throw immediately with context. Let errors propagate to boundaries where they can be logged and translated to user-appropriate responses. Never swallow. Never return defaults to hide failure.