Official Python SDK for the Memic Context Engineering API.
Upload documents, process them into searchable chunks, and perform semantic search with metadata filters.
pip install memicfrom memic import Memic, MetadataFilters, PageRange
# Initialize client (uses MEMIC_API_KEY env var if not provided)
client = Memic(api_key="mk_...")
# List projects
projects = client.list_projects()
for p in projects:
print(f"{p.name}: {p.id}")
# Upload a file (waits for processing by default)
file = client.upload_file(
project_id="your-project-id",
file_path="/path/to/document.pdf",
reference_id="lesson_123" # Optional: for linking with external systems
)
print(f"Uploaded: {file.id}, status: {file.status}")
# Search with filters
results = client.search(
query="key findings about climate change",
project_id="your-project-id",
top_k=10,
min_score=0.7,
filters=MetadataFilters(
reference_id="TG_G1_Math",
page_range=PageRange(gte=1, lte=50)
)
)
for result in results:
print(f"[{result.score:.2f}] {result.file_name} (p.{result.page_number})")
print(f" {result.content[:200]}...")- File Upload: 3-step presigned URL flow for efficient uploads
- Wait for Ready: Automatic polling until file processing completes
- Semantic Search: Vector similarity search with rich metadata
- Metadata Filters: Filter by reference_id, page numbers, categories
- Simple API: Single client class, no complex patterns
Set your API key via environment variable or constructor:
export MEMIC_API_KEY=mk_your_api_key_here# Or pass directly
client = Memic(api_key="mk_...")For development or self-hosted deployments:
client = Memic(
api_key="mk_...",
base_url="https://your-api.example.com"
)client = Memic(
api_key: str = None, # Uses MEMIC_API_KEY env var if not provided
base_url: str = None, # Default: https://api.memic.ai
timeout: int = 30 # Request timeout in seconds
)List all projects in your organization.
Upload a file to a project.
| Parameter | Type | Default | Description |
|---|---|---|---|
project_id |
str | required | Target project ID |
file_path |
str/Path | required | Path to file |
wait_for_ready |
bool | True | Wait for processing to complete |
reference_id |
str | None | External reference ID |
metadata |
dict | None | Custom metadata |
poll_interval |
float | 2.0 | Seconds between status checks |
poll_timeout |
float | 300 | Max wait time in seconds |
Get current processing status of a file.
Search for content across documents.
| Parameter | Type | Default | Description |
|---|---|---|---|
query |
str | required | Search query |
project_id |
str | None | Limit to project |
file_ids |
List[str] | None | Limit to specific files |
top_k |
int | 10 | Number of results |
min_score |
float | 0.7 | Minimum similarity score |
filters |
MetadataFilters | None | Metadata filters |
Enum with processing states:
UPLOADING,UPLOADED,UPLOAD_FAILEDCONVERSION_STARTED,CONVERSION_COMPLETE,CONVERSION_FAILEDPARSING_STARTED,PARSING_COMPLETE,PARSING_FAILEDCHUNKING_STARTED,CHUNKING_COMPLETE,CHUNKING_FAILEDEMBEDDING_STARTED,EMBEDDING_COMPLETE,EMBEDDING_FAILEDREADY
Properties:
.is_failed- True if status indicates failure.is_processing- True if still processing
MetadataFilters(
reference_id: str = None, # Filter by reference ID
reference_ids: List[str] = None, # Multiple reference IDs (OR)
page_number: int = None, # Exact page match
page_numbers: List[int] = None, # Multiple pages (OR)
page_range: PageRange = None, # Page range
category: str = None, # Category filter
document_type: str = None # Document type filter
)from memic import MemicError, AuthenticationError, NotFoundError, APIError
try:
results = client.search(query="test")
except AuthenticationError:
print("Invalid API key")
except NotFoundError:
print("Resource not found")
except APIError as e:
print(f"API error: {e.status_code} - {e.message}")
except MemicError as e:
print(f"Error: {e.message}")# Install dev dependencies
pip install -e ".[dev]"
# Run tests
pytest tests/
# Type check
mypy src/memic/
# Lint
ruff check src/MIT License - see LICENSE for details.