bundled-skills/sdk-dx/SKILL.md
Design SDKs that developers love to use—APIs that feel native, error messages that guide, and experiences that reduce friction. This skill covers creating SDKs that drive adoption through exceptional developer experience rather than aggressive marketing. Trigger phrases: "SDK design",...
npx skillsauth add FrancoStino/opencode-skills-antigravity sdk-dxInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Use this skill when you need design SDKs that developers love to use—APIs that feel native, error messages that guide, and experiences that reduce friction. This skill covers creating SDKs that drive adoption through exceptional developer experience rather than aggressive marketing. Trigger phrases: "SDK design",...
The best SDK marketing is an SDK that developers can't stop talking about. When your SDK makes developers feel productive and competent, they become your advocates. When it frustrates them, no amount of marketing will save you.
SDK developer experience (DX) encompasses everything a developer feels when using your library:
Great SDK DX is a competitive advantage. Developers choose tools that make them feel smart.
Review the developer-audience-context skill to understand:
SDK design decisions should flow from deep understanding of your users.
The most frequent use case should require the least code.
Good Design:
# Common case: send a simple message
client.messages.send("Hello world", to="+1234567890")
# Full control when needed
client.messages.send(
body="Hello world",
to="+1234567890",
from_="+0987654321",
status_callback="https://...",
media_urls=["https://..."]
)
Bad Design:
# Every call requires full configuration
message = Message(
body="Hello world",
to=PhoneNumber("+1234567890"),
from_=PhoneNumber(config.get_default_from()),
options=MessageOptions(
status_callback=None,
media_urls=[]
)
)
client.messages.send(message)
Start simple, reveal complexity as needed.
// Level 1: Simplest possible usage
const result = await client.analyze("Hello world");
// Level 2: Common options
const result = await client.analyze("Hello world", {
language: "en",
features: ["sentiment", "entities"]
});
// Level 3: Full control
const result = await client.analyze("Hello world", {
language: "en",
features: ["sentiment", "entities"],
model: "v2-large",
timeout: 30000,
retries: { max: 3, backoff: "exponential" }
});
Catch errors as early as possible, with actionable messages.
Good:
# Validation at construction time
client = MyClient(api_key="")
# Raises immediately: ValueError: API key cannot be empty.
# Get your API key at https://dashboard.example.com/keys
# Clear error at runtime
client.users.get("invalid-id")
# Raises: NotFoundError: User 'invalid-id' not found.
# Use client.users.list() to see available users.
Bad:
client = MyClient(api_key="") # No validation
result = client.users.get("invalid-id")
# Returns: None (is this an error? empty result? who knows?)
# Or worse: raises generic Exception with stack trace
Default values should work for most cases without configuration.
// This should just work without configuration
const client = new MyClient({ apiKey: process.env.MY_API_KEY });
// Sensible defaults:
// - Automatic retries with exponential backoff
// - Reasonable timeouts
// - JSON content type
// - Standard auth headers
// - Connection pooling
Error messages are documentation. Make them helpful.
Every error message should answer:
Good:
AuthenticationError: Invalid API key provided.
The API key 'sk_test_abc...' (test key) cannot be used for
production requests.
To fix this:
1. Go to https://dashboard.example.com/keys
2. Copy your production API key (starts with 'sk_live_')
3. Update your environment variable: MY_API_KEY=sk_live_...
Docs: https://docs.example.com/authentication
Bad:
Error: 401 Unauthorized
Create specific error types that developers can catch:
from myapi.errors import (
AuthenticationError, # Invalid/missing credentials
AuthorizationError, # Valid creds, insufficient permissions
ValidationError, # Invalid input data
NotFoundError, # Resource doesn't exist
RateLimitError, # Too many requests
ServerError, # Our fault, retry might help
)
try:
client.users.get(user_id)
except NotFoundError as e:
# Handle missing user specifically
except AuthenticationError as e:
# Handle auth issues specifically
except MyAPIError as e:
# Catch-all for other API errors
// Bad: generic error
throw new Error("Invalid parameter");
// Good: contextual error
throw new ValidationError({
message: "Invalid phone number format",
field: "to",
value: "+1abc",
expected: "E.164 format (e.g., +14155551234)",
docs: "https://docs.example.com/phone-numbers"
});
Type safety is documentation that never goes stale.
// Define explicit types for all inputs and outputs
interface User {
id: string;
email: string;
name: string;
createdAt: Date;
metadata?: Record<string, unknown>;
}
interface CreateUserInput {
email: string;
name: string;
metadata?: Record<string, unknown>;
}
// Return types are explicit
async function createUser(input: CreateUserInput): Promise<User> {
// ...
}
// Use discriminated unions for responses
type ApiResponse<T> =
| { success: true; data: T }
| { success: false; error: ApiError };
Design for IDE autocomplete:
// Good: autocomplete shows all options
client.messages.create({
to: "+1...", // IDE shows: (property) to: string
body: "...", // IDE shows: (property) body: string
// User types 'me' and sees 'mediaUrls' autocomplete
});
// Bad: requires memorization
client.send("messages", { /* what goes here? */ });
// Good: constrained values with autocomplete
type MessageStatus = "queued" | "sending" | "sent" | "failed";
interface Message {
status: MessageStatus; // IDE shows valid values
}
// Bad: any string accepted
interface Message {
status: string; // No guidance, errors at runtime
}
Structure your SDK so IDE features help developers:
// Namespace methods logically
client.users.get(id)
client.users.list()
client.users.create(data)
client.users.update(id, data)
client.users.delete(id)
// After typing 'client.users.' the IDE shows all user operations
/**
* Creates a new user in your organization.
*
* @param input - The user details
* @param input.email - Must be a valid email address
* @param input.name - Display name (max 100 characters)
* @returns The created user with generated ID
* @throws {ValidationError} If email format is invalid
* @throws {ConflictError} If email already exists
*
* @example
* const user = await client.users.create({
* email: "[email protected]",
* name: "Jane Developer"
* });
*/
async createUser(input: CreateUserInput): Promise<User>
def send_message(self, body: str, to: str, **kwargs) -> Message:
"""
Send an SMS message.
Args:
body: The message content (max 1600 characters)
to: Recipient phone number in E.164 format
Returns:
Message object with ID and status
Example:
>>> message = client.messages.send(
... body="Hello from Python!",
... to="+14155551234"
... )
>>> print(message.status)
'queued'
"""
Follow semver strictly:
Breaking changes (require major version bump):
Not breaking (minor or patch):
import warnings
def old_method(self):
"""
.. deprecated:: 2.3.0
Use :meth:`new_method` instead. Will be removed in 3.0.0.
"""
warnings.warn(
"old_method() is deprecated, use new_method() instead. "
"See migration guide: https://docs.example.com/migrate-v3",
DeprecationWarning,
stacklevel=2
)
return self.new_method()
# Migrating from v2 to v3
## Overview
Version 3 introduces [major change] and removes [deprecated feature].
Migration typically takes [time estimate].
## Breaking Changes
### 1. Client Initialization
**Before (v2):**
```python
client = MyClient(key="...")
After (v3):
client = MyClient(api_key="...")
Why: Consistency with other SDK parameters.
...
client.old_method() - Use client.new_method() insteadLegacyClass - Use ModernClass instead
### Codemods and Automation
When possible, provide automated migration:
```bash
# Provide migration scripts
npx @myapi/migrate-v3
# Or codemods
npx jscodeshift -t @myapi/codemods/v2-to-v3 src/
Python: Use snake_case, context managers, generators
# Pythonic
with client.batch() as batch:
for user in client.users.list():
batch.add(user.send_notification("Hello"))
# Not Pythonic
users = client.getUsers()
batch = client.createBatch()
for i in range(len(users)):
batch.addOperation(users[i].sendNotification("Hello"))
batch.execute()
JavaScript: Use Promises, async/await, destructuring
// Idiomatic JS
const { data, error } = await client.users.get(id);
// Not idiomatic
client.users.get(id, function(err, result) {
if (err) { /* callback hell */ }
});
Go: Use error returns, interfaces, channels
// Idiomatic Go
user, err := client.Users.Get(ctx, userID)
if err != nil {
return fmt.Errorf("getting user: %w", err)
}
// Not idiomatic
user := client.Users.Get(userID) // panics on error
tools
Authorized security assessment of LLM applications and AI agents: prompt injection, tool abuse, RAG exposure, memory poisoning, system-prompt extraction, and agent-compliance engineering per OWASP LLM/ASI Top 10.
development
Builds two parameterized UI modes—流光溢彩白 (iridescent white) and 五彩斑斓黑 (colorful black)—with OKLCH, WebGL/CSS fallback, vision gating, screenshot QA, and total/per-color intensity reports. Use when a UI request names either mode or needs measured color parameters.
tools
Delegate coding tasks to the Kimi Code CLI (`kimi`) only when the user explicitly requests it, while the orchestrator retains review and landing responsibility.
development
Front-end JavaScript reverse engineering: locate signature chains, analyze encrypted request parameters, sample runtime behavior, and reproduce logic locally in Node for evidence-based output.