Quickstart Guide
Get up and running with Skimly in minutes. Choose your preferred language or use raw HTTP - both approaches are fully supported.
Our v2.0 SDKs provide streaming support, tool calling,TypeScript types, and Anthropic-compatible interfaceswith automatic compression and error handling.
1. Installation
Choose your preferred language and install the appropriate package:
npm install @skimly/sdk@^2.0.0
2. Authentication
Set your API key as an environment variable. Both SDKs will automatically read this:
export SKIMLY_KEY="YOUR_API_KEY" # For production, also set: export SKIMLY_BASE="https://api.skimly.dev"
Before you can use Skimly, you need to create an API key:
- Visit the API Keys page in your dashboard
- Click "Create Key" and choose "Test" mode for development
- Copy the generated key (starts with
sk_test_) - Set it as your
SKIMLY_KEYenvironment variable
New to Skimly? Start with the Onboarding guide for a complete walkthrough.
3. Create a Blob
Upload large, rarely-changing content (like policies, documentation, or long conversation threads) to get a blob ID that you can reference in future requests.
// npm i @skimly/sdk@^2.0.0
import { SkimlyClient } from '@skimly/sdk'
const client = new SkimlyClient({
apiKey: process.env.SKIMLY_KEY!,
baseURL: 'http://localhost:8000'
})
const blob = await client.createBlob('big content...', 'text/plain')
console.log('Blob ID:', blob.blob_id)
// Idempotent upload (avoids duplicates)
const dedupedBlob = await client.createBlobIfNew('big content...', 'text/plain')
console.log('Deduped Blob ID:', dedupedBlob.blob_id)Use createBlobIfNew() (Node.js) or create_blob_if_new() (Python) to automatically avoid re-uploading identical content during the same process.
4. Chat with Blob Reference
Use the blob ID as a lightweight pointer in your chat requests instead of sending the full content. This dramatically reduces token usage while maintaining full context.
// npm i @skimly/sdk@^2.0.0
import { SkimlyClient } from '@skimly/sdk'
const client = new SkimlyClient({
apiKey: process.env.SKIMLY_KEY!,
baseURL: 'http://localhost:8000'
})
// Anthropic-style interface with compression
const message = await client.messages.create({
provider: 'openai',
model: 'gpt-4o-mini',
max_tokens: 1024,
messages: [{ role: 'user', content: 'Hello from Node!' }]
})
console.log(message.content[0].text)
console.log('Tokens saved:', message.skimly_meta?.tokens_saved)5. Switch Providers with One Line
One of the biggest benefits of Skimly is how easy it is to switch between providers. Just change the provider field and you're done!
OpenAI
# Node.js
await client.messages.create({
provider: 'openai',
model: 'gpt-4o-mini',
max_tokens: 1024,
messages: [{ role: 'user', content: 'Hello!' }]
})
# Python
await client.messages.create({
'provider': 'openai',
'model': 'gpt-4o-mini',
'max_tokens': 1024,
'messages': [{'role': 'user', 'content': 'Hello!'}]
})Anthropic
# Node.js - Anthropic-compatible interface
await client.messages.create({
provider: 'anthropic',
model: 'claude-3-5-sonnet-20241022',
max_tokens: 1024,
messages: [{ role: 'user', content: 'Hello!' }]
})
# Python
await client.messages.create({
'provider': 'anthropic',
'model': 'claude-3-5-sonnet-20241022',
'max_tokens': 1024,
'messages': [{'role': 'user', 'content': 'Hello!'}]
})6. Enable Smart Compression
Skimly automatically applies intelligent compression based on content analysis. Use mode: "compact" to enable advanced compression features.
Smart Compression Example
# Node.js
const response = await client.messages.create({
provider: 'openai',
model: 'gpt-4o-mini',
max_tokens: 2048,
messages: [
{ role: 'user', content: 'Analyze this error log: b_abc123' }
]
})
console.log('Tokens saved:', response.skimly_meta?.tokens_saved)
# Python
response = await client.messages.create({
'provider': 'openai',
'model': 'gpt-4o-mini',
'max_tokens': 2048,
'messages': [
{'role': 'user', 'content': 'Analyze this error log: b_abc123'}
]
})
print('Tokens saved:', response['skimly_meta']['tokens_saved'])Skimly automatically compresses verbose AI responses while preserving the ability to expand to full content when needed. Smart Truncation provides up to 99% token savings.
7. Real-time Streaming
Stream AI responses in real-time using Server-Sent Events. Both SDKs provide full streaming support with proper chunking and error handling.
Streaming Example
# Node.js
const stream = client.messages.stream({
provider: 'anthropic',
model: 'claude-3-5-sonnet-20241022',
max_tokens: 1024,
messages: [{ role: 'user', content: 'Write a story' }],
stream: true
})
for await (const chunk of stream) {
if (chunk.type === 'content_block_delta' && chunk.delta?.text) {
process.stdout.write(chunk.delta.text)
}
}
# Python
async for chunk in client.messages.stream({
'provider': 'anthropic',
'model': 'claude-3-5-sonnet-20241022',
'max_tokens': 1024,
'messages': [{'role': 'user', 'content': 'Write a story'}],
'stream': True
}):
if chunk['type'] == 'content_block_delta' and chunk.get('delta', {}).get('text'):
print(chunk['delta']['text'], end='', flush=True)8. Advanced Features
Content Transformation
Use the transform endpoint to compress tool results with context awareness.
# Node.js
await client.transform({
result: "Large tool output...",
tool_name: "bash",
command: "npm run build"
})
# Python
await client.transform(
result="Large tool output...",
tool_name="bash",
command="npm run build"
)Range Reads
Efficiently read portions of large blobs without downloading everything.
# Node.js
const content = await client.fetchBlob('b_abc123', { start: 0, end: 1024 })
# Python
content = await client.fetch_blob('b_abc123', start=0, end=1024)9. Next Steps
Use our official SDKs for the best developer experience, or integrate directly with the HTTP endpoints. Both approaches provide the same functionality.