Python 3.13 — Fast

Python 3.13 — Fast starts executing immediately — there is no container to spin up between calls. It trades full PyPI availability for the lowest-latency Python execution on the platform.

Choose this runtime when your dependencies are pure-Python or from the supported package set and you want the fastest possible response time.

---

Handler signature

Your main file must be named cortexone_function.py and must define a function called cortexone_handler:

import json

def cortexone_handler(event, context):
    result = {"message": "Hello, World!"}
    return {"statusCode": 200, "body": json.dumps(result)}

Parameter	Type	Description
event	dict	The input payload sent by the caller
context	object	Execution metadata (fn_id, user_id)

The handler returns a dict with statusCode and body. Use json.dumps() for the body.

---

The event object

event is whatever JSON the caller passes:

import json

def cortexone_handler(event, context):
    name = event["name"]
    numbers = event["numbers"]
    return {
        "statusCode": 200,
        "body": json.dumps({"greeting": f"Hello, {name}!", "total": sum(numbers)})
    }

---

Multiple files

You can upload multiple files. The main entry file must be cortexone_function.py containing cortexone_handler. Additional .py files are importable with standard imports.

Supported file types: .py, .csv, .md, .txt, .json, .yaml, .yml.

helpers.py

def greet(name):
    return f"Hello, {name}!"

cortexone_function.py

import json
from helpers import greet

def cortexone_handler(event, context):
    return {"statusCode": 200, "body": json.dumps({"message": greet(event["name"])})}

You can also upload data files and read them with open():

import json

def cortexone_handler(event, context):
    with open("/data/config.json") as f:
        cfg = json.load(f)
    return {"statusCode": 200, "body": json.dumps(cfg)}

---

Dependencies

requirements.txt

Add a requirements.txt alongside cortexone_function.py:

numpy
pandas==2.2.0
scipy>=1.12

• Standard version specifiers are supported (==, >=, !=, ~=, etc.)
• URL-based dependencies (lines starting with https://) are not supported
• Extras like pandas[excel] are accepted but the extra itself is ignored

Supported packages

283 packages are pre-built and ready to use. See the full list below. If a package you need isn’t there, contact support and we’ll add it.

If a package is unavailable, the error will say so. You can also switch to the standard Python 3.13 runtime to access the full PyPI ecosystem.

---

Secrets

Secrets are accessed through process["env"] — not os.environ:

import json

def cortexone_handler(event, context):
    api_key = process["env"].get("MY_API_KEY", "")
    return {"statusCode": 200, "body": json.dumps({"ready": len(api_key) > 0})}

process["env"] is always available without any import. Store credentials as Secrets at /user/secrets and attach them to your tool from the Code tab in Studio. See Secrets for the full workflow.

---

Built-in utilities

These functions are always available without any import:

Function	Description
hash_digest(algorithm, data)	Hash a string — "sha256", "sha512", "sha1", "md5"
btoa(s)	Base64-encode a string
atob(s)	Base64-decode a string
crypto.random_uuid()	Generate a random UUID v4 string
crypto.get_random_values(n)	Return n random bytes as a list of integers (max 65536)
json_encode(obj)	Serialize to a compact JSON string
json_decode(s)	Parse a JSON string
text_encode(s)	Encode a string to UTF-8 bytes (list of ints)
text_decode(byte_list)	Decode a UTF-8 byte list to a string
url_parse(url)	Parse a URL into components (scheme, host, path, params, etc.)
structured_clone(obj)	Deep-copy a dict or list
fetch(url, options)	Synchronous HTTP request. options keys: method, headers, body. Returns {status, ok, text, json, headers}.

import json

def cortexone_handler(event, context):
    text = event.get("text", "")
    return {
        "statusCode": 200,
        "body": json.dumps({
            "sha256": hash_digest("sha256", text),
            "base64": btoa(text),
            "id": crypto.random_uuid(),
        })
    }

---

What is available

Feature	Available
Python 3.13 standard library	Yes — math, json, re, datetime, collections, itertools, hashlib, csv, os.path, and more
process["env"]	Yes — read your Secrets
File I/O (open, read, write)	Yes — files are temporary and scoped to the current execution
Built-in utilities	Yes — hash_digest, btoa/atob, crypto, json_encode/json_decode, etc.
fetch(url, options)	Yes — synchronous HTTP (see fetch builtin)
subprocess / multiprocessing	No — blocked
shutil, venv, pip	No — blocked
Dangerous os operations	No — os.system, os.popen, os.fork, os.kill, etc. are blocked
os.environ	No — use process["env"] instead

---

Resource limits

Limit	Default
Max memory	512 MB
Max runtime	300 seconds
CPU limit	2.0 cores

---

Example: Number processing with numpy

import numpy as np
import json

def cortexone_handler(event, context):
    arr = np.array(event["numbers"])
    return {
        "statusCode": 200,
        "body": json.dumps({
            "mean": float(arr.mean()),
            "sum": int(arr.sum()),
            "max": float(arr.max()),
        })
    }

---

Example: DataFrame analysis with pandas

import pandas as pd
import json

def cortexone_handler(event, context):
    df = pd.DataFrame(event["rows"])
    return {
        "statusCode": 200,
        "body": json.dumps({
            "shape": list(df.shape),
            "columns": list(df.columns),
            "mean": df.select_dtypes("number").mean().to_dict(),
        })
    }

---

Error responses

If your handler raises an exception, the platform catches it and returns:

{
  "statusCode": 500,
  "body": "{\"error\": \"Script execution failed\", \"type\": \"ExecutionError\", \"details\": \"...\"}"
}

Timeouts return statusCode: 408. Memory limit exceeded returns statusCode: 507.

---

Notes

• The main file must be named cortexone_function.py and must contain cortexone_handler.
• Each execution gets a completely fresh Python namespace — no state is shared between calls.
• Files written during one execution are not available in the next.
• Installed packages are cached for the lifetime of the worker, so repeated calls with the same packages are fast.
• For outbound HTTP or packages outside the supported set, use the standard Python 3.13 runtime instead.

---

Available packages

The following 283 packages are pre-built and available to install via requirements.txt. Install them by name — no version pinning required unless you need a specific release.

If you need a package that isn’t listed here, contact us through the Support option in the app and we’ll add it for you.

Package	Version	Description
affine	2.4.0	Affine transformation matrices for 2D coordinate transforms
aiohappyeyeballs	2.6.1	Happy Eyeballs algorithm for async TCP connections
aiohttp	3.13.5	Async HTTP client/server framework
aiosignal	1.4.0	Signal/callback list for asyncio
altair	6.0.0	Declarative statistical visualization library
annotated-doc	0.0.4	Annotation and documentation utilities
annotated-types	0.7.0	Type annotation metadata for validation
anyio	4.13.0	Async I/O compatibility layer (asyncio/trio)
apsw	3.51.3.0	Another Python SQLite Wrapper — full SQLite API
argon2-cffi	23.1.0	Argon2 password hashing library
argon2-cffi-bindings	25.1.0	Low-level CFFI bindings for Argon2
astropy	7.2.0	Astronomy and astrophysics toolkit
astropy_iers_data	0.2026.4.1.15.5.49	IERS Earth rotation and time data for Astropy
asttokens	3.0.1	Maps AST nodes to source code tokens
async-timeout	5.0.1	Asyncio-compatible timeout context manager
atomicwrites	1.4.1	Atomic file writing on POSIX systems
attrs	26.1.0	Boilerplate-free class definitions with validation
audioop-lts	0.2.2	Audio operations (Python 3.13 LTS backport)
awkward-cpp	52	C++ backend for Awkward Array (ragged arrays)
b2d	0.7.4	Box2D 2D rigid body physics engine bindings
bcrypt	5.0.0	bcrypt password hashing
beautifulsoup4	4.14.3	HTML and XML parsing and scraping
bilby.cython	0.5.4	Cython extensions for Bilby Bayesian inference
biopython	1.87	Computational biology and bioinformatics toolkit
bitarray	3.8.1	Efficient bit array implementation
bitstring	4.4.0	Bit-level data creation, parsing and manipulation
bleach	6.3.0	HTML sanitization and text linkification
blosc2	4.1.2	High-performance compressed array storage
bokeh	3.9.0	Interactive browser-based visualization
boost-histogram	1.7.1	High-performance histogram filling (Boost.Histogram)
Bottleneck	1.6.0	Fast NumPy array functions (NaN-aware, Cython)
brotli	1.2.0	Brotli compression and decompression
cachetools	7.0.5	Memoization and caching decorators
Cartopy	0.25.0	Cartographic projections and geospatial visualization
casadi	3.7.2	Symbolic framework for numerical optimization
cbor-diag	1.1.2	CBOR (Concise Binary Object Representation) diagnostics
certifi	2026.4.22	Mozilla CA certificate bundle
cffi	2.0.0	C Foreign Function Interface for Python
cftime	1.6.5	Time-handling for netCDF and CF conventions
charset-normalizer	3.4.7	Character encoding detection
clarabel	0.11.1	Interior-point conic optimization solver
click	8.3.1	Composable command-line interface toolkit
cligj	0.7.2	Click parameter types for GeoJSON
clingo	5.8.0	Answer Set Programming grounder and solver
cloudpickle	3.1.2	Extended pickling for closures and lambdas
cmyt	2.0.2	Scientific colormaps from yt
cobs	1.2.2	Consistent Overhead Byte Stuffing encoding
colorspacious	1.1.2	Color space conversions and perceptual distance
contourpy	1.3.3	Contour line and filled contour computation
coolprop	7.2.0	Thermodynamic and transport properties of fluids
coverage	7.13.5	Code coverage measurement
crc32c	2.8	CRC32C checksum (hardware-accelerated)
crcmod	1.7	Cyclic redundancy check (CRC) calculation
cryptography	47.0.0	Cryptographic primitives and recipes
cssselect	1.4.0	CSS selector parsing and translation to XPath
cvxpy-base	1.8.2	Convex optimization problem modeling
cycler	0.12.1	Composable cycles for matplotlib style parameters
cysignals	1.12.3	Unix signal handling for Cython/C extensions
cytoolz	1.1.0	Cython-accelerated functional utilities (toolz)
decorator	5.2.1	Signature-preserving function decorators
demes	0.2.3	Demographic history model specification
deprecated	1.3.1	Mark functions/classes as deprecated with warnings
deprecation	2.1.0	Deprecation helpers and warnings
diskcache	5.6.3	Disk-backed cache and deque
distlib	0.4.0	Low-level Python distribution utilities
distro	1.9.0	Linux OS distribution detection
dnspython	2.8.0	DNS toolkit — queries, zones, and DNSSEC
docutils	0.22.4	reStructuredText document processing
donfig	0.8.1.post1	Configuration management for libraries
duckdb	1.5.1	In-process analytical SQL database engine
ewah_bool_utils	1.3.0	EWAH bitmap compression for yt
exceptiongroup	1.3.1	PEP 654 exception groups (backport)
executing	2.2.1	Inspect code being executed at runtime
fastapi	0.136.1	High-performance async web API framework
fastcan	0.5.0	Fast canonical correlation analysis
fiona	1.9.5	Pythonic API for OGR vector geospatial data
fonttools	4.62.1	Read, write and manipulate font files
freesasa	2.2.1	Solvent-accessible surface area of proteins
frozenlist	1.8.0	List that can be made immutable
fsspec	2026.3.0	Filesystem-spec — unified interface to filesystems
future	1.0.0	Python 2/3 compatibility bridge
galpy	1.11.2	Galactic dynamics: orbits, potentials, action-angle
geopandas	1.1.3	Geospatial extensions for pandas
gmpy2	2.3.0	Multiple-precision arithmetic (GMP/MPFR/MPC)
google-crc32c	1.8.0	CRC32C checksums using Google’s implementation
gsw	3.6.21	TEOS-10 Gibbs SeaWater oceanographic toolbox
h11	0.16.0	Pure-Python HTTP/1.1 implementation
h3	4.4.2	Uber H3 hexagonal hierarchical geospatial indexing
h5py	3.13.0	HDF5 binary data format interface
highspy	1.13.1	Python bindings for HiGHS linear/integer optimizer
html5lib	1.1	Standards-compliant HTML parser
httpcore	1.0.9	Minimal HTTP client library
httpx	0.28.1	Fully featured async/sync HTTP client
idna	3.11	Internationalized Domain Names in Applications (IDNA)
igraph	1.0.0	Network analysis and graph algorithms
imageio	2.37.3	Read and write images, video, and volumetric data
iminuit	2.30.1	Function minimization and fitting (MINUIT)
iniconfig	2.3.0	Simple INI-file configuration parser
ipython	9.12.0	Enhanced interactive Python shell
jedi	0.19.2	Autocompletion and static analysis for Python
Jinja2	3.1.6	Templating engine for Python
jiter	0.13.0	Fast iterable JSON parser
joblib	1.5.3	Lightweight pipelining and parallelism
jsonpatch	1.33	JSON Patch (RFC 6902) apply and generate
jsonpointer	3.1.1	JSON Pointer (RFC 6901) resolution
jsonschema	4.26.0	JSON Schema validation
jsonschema_specifications	2025.9.1	JSON Schema specification meta-schemas
kiwisolver	1.5.0	Cassowary constraint solving algorithm
lakers-python	0.6.2	LAKERS OSCORE+EDHOC protocol implementation
lazy_loader	0.5	Lazy module attribute loading
lazy-object-proxy	1.12.0	Lazy initialization of objects
libcst	1.8.6	Concrete syntax tree for Python source code
lightgbm	4.6.0	Gradient boosting framework (LightGBM)
logbook	1.9.2	Logging system for Python
lxml	6.0.2	Feature-rich XML and HTML processing
lz4	4.4.5	LZ4 fast compression/decompression
MarkupSafe	3.0.3	Safely handle text with HTML/XML special characters
matplotlib	3.10.8	2D plotting and visualization
matplotlib-inline	0.2.1	Matplotlib backend for inline display
memory-allocator	0.2.0	Custom memory allocation utilities
ml_dtypes	0.5.4	NumPy dtypes for machine learning (bfloat16, float8, etc.)
mmh3	5.2.1	MurmurHash3 non-cryptographic hash
more-itertools	11.0.1	Extended itertools — additional iterator utilities
mpmath	1.4.1	Arbitrary-precision floating-point arithmetic
msgpack	1.1.2	Efficient binary serialization (MessagePack)
msgspec	0.20.0	High-performance JSON/MessagePack serialization with type validation
msprime	1.4.1	Ancestral recombination graph simulation
multidict	6.7.1	Multivalue dictionary (for HTTP headers etc.)
munch	4.0.0	Dot-access dictionary
mypy	1.19.1	Static type checker for Python
narwhals	2.18.1	Dataframe-agnostic API (pandas, polars, modin, etc.)
ndindex	1.10.1	NumPy-like index types with safe operations
netcdf4	1.7.4	Read/write NetCDF-4 and NetCDF-3 files
networkx	3.6.1	Network creation, analysis and algorithms
newick	1.11.0	Newick phylogenetic tree format parser
nh3	0.3.4	HTML sanitization (Rust-based, ammonia)
nlopt	2.9.1	Nonlinear optimization (many algorithms)
nltk	3.9.4	Natural language processing toolkit
numcodecs	0.15.1	Array compression codecs for Zarr and NumPy
numpy	2.4.3	N-dimensional array computing
openai	2.30.0	OpenAI API client
opencv-python	4.11.0.86	Computer vision and image processing (OpenCV)
optlang	1.9.0	Mathematical programming language for optimization
orjson	3.11.8	Fast JSON serialization/deserialization
packaging	26.1	Package version parsing and specifiers
pandas	3.0.2	Data manipulation and analysis
parso	0.8.6	Python parser with error recovery
patsy	1.0.2	Statistical model formula notation (R-style)
pcodec	1.0.1	Lossless compression for numeric data
peewee	4.0.4	Lightweight ORM for SQLite, MySQL, PostgreSQL
phispy	5.0.6	Prophage identification in bacterial genomes
pi-heif	1.3.0	HEIF/HEIC image format read/write
Pillow	12.2.0	Image processing (PIL fork)
pillow-heif	1.3.0	HEIF/HEIC plugin for Pillow
pkgconfig	1.6.0	Interface to pkg-config
platformdirs	4.9.4	Platform-specific user/app directories
pluggy	1.6.0	Plugin management and hook system
ply	3.11	Python Lex-Yacc parser tools
polars	1.33.1	High-performance DataFrame library (Rust-based)
prompt_toolkit	3.0.52	Building interactive command-line applications
propcache	0.4.1	Property caching decorator
protobuf	7.34.1	Protocol Buffers serialization
pure-eval	0.2.3	Safely evaluate simple Python expressions
py	1.11.0	Python testing and code utilities
pyarrow	22.0.0	Apache Arrow columnar in-memory data
pyclipper	1.4.0	Polygon clipping and offsetting (Clipper)
pycparser	3.0	Complete C99 parser in Python
pycryptodome	3.23.0	Self-contained cryptographic algorithms
pydantic	2.12.5	Data validation using Python type hints
pydantic_core	2.41.5	Core validation engine for Pydantic
pyerfa	2.0.1.5	Python bindings for ERFA astronomical library
pygame-ce	2.5.7	Game development framework (community edition)
Pygments	2.20.0	Syntax highlighting for 300+ languages
pyheif	0.8.0	Read HEIF/HEIC image files
pyiceberg	0.11.1	Apache Iceberg table format client
pyinstrument	5.1.2	Statistical call-stack profiler
pymongo	4.16.0	MongoDB driver for Python
PyMuPDF	1.27.2.2	PDF and XPS document processing (MuPDF)
pynacl	1.6.2	Python bindings to libsodium (NaCl cryptography)
pyparsing	3.3.2	PEG parsing expression grammar library
pyproj	3.7.2	Cartographic projections (PROJ library)
pyroaring	1.0.4	Compressed bitmap sets (Roaring Bitmaps)
pyrodigal	3.7.1	Gene prediction (Prodigal) Python interface
pyrsistent	0.20.0	Immutable persistent data structures
pysam	0.23.0	SAM/BAM/CRAM/VCF/BCF genomics file access
pyshp	3.0.3	Read/write ESRI Shapefile format
pytaglib	3.2.0	Audio file metadata reading/writing (TagLib)
pytest	9.0.2	Testing framework
pytest-asyncio	0.25.3	Asyncio support for pytest
pytest-benchmark	4.0.0	Benchmarking fixtures for pytest
pytest_httpx	0.36.0	Mock httpx requests in pytest
python-calamine	0.6.2	Excel/ODS file reading (Calamine, Rust)
python-dateutil	2.9.0.post0	Powerful extensions to datetime
python-flint	0.8.0	Arbitrary-precision number theory (FLINT)
python-flirt	0.9.10	Fast Library Identification and Recognition Technology
python-sat	1.8.dev26	SAT solver toolkit (pysat)
python-solvespace	3.0.8	Parametric CAD constraint solver
pytz	2026.1.post1	IANA timezone database for Python
pywavelets	1.9.0	Discrete wavelet transforms
pyxirr	0.10.8	Financial functions — IRR, XIRR, NPV, etc.
pyyaml	6.0.3	YAML parser and emitter
rasterio	1.5.0	Geospatial raster data I/O (GDAL)
rateslib	2.7.1	Fixed income and derivatives pricing
rebound	4.4.7	N-body gravitational dynamics simulator
reboundx	4.4.1	Additional forces and effects for REBOUND
referencing	0.37.0	JSON reference resolution
regex	2026.3.32	Alternative re module with additional features
requests	2.33.1	HTTP library for humans
retrying	1.4.2	General retry decorator
rich	14.3.3	Rich text, tables and progress bars in terminal
RobotRaconteur	1.2.7	Distributed robotics middleware
rpds-py	0.30.0	Immutable persistent data structures (Rust)
ruamel.yaml	0.19.1	YAML 1.2 parser with round-trip preservation
safetensors	0.7.0	Safe tensor serialization format (Hugging Face)
scikit-image	0.25.2	Image processing algorithms
scikit-learn	1.8.0	Machine learning library
scipy	1.17.1	Scientific computing — optimization, integration, signal processing
screed	1.1.3	DNA/RNA sequence file reading
sentencepiece	0.2.1	Unsupervised text tokenizer (BPE/Unigram)
setuptools	82.0.1	Package building and distribution utilities
shapely	2.1.2	Geometric objects, predicates and operations
simplejson	3.20.2	JSON encoder/decoder
sisl	0.16.4	Electronic structure and transport library
six	1.17.0	Python 2 and 3 compatibility utilities
smart-open	7.5.1	Transparent open() for S3, GCS, HDFS and local
sniffio	1.3.1	Detect which async library is running
sortedcontainers	2.4.0	Sorted list, dict and set (pure Python, fast)
soundfile	0.12.1	Read and write sound files (libsndfile)
soupsieve	2.8.3	CSS selector implementation for BeautifulSoup
sourmash	4.8.14	Sequence sketching and genome comparison
soxr	0.5.0.post1	High-quality audio resampling
sparseqr	1.2	Sparse QR decomposition (SuiteSparseQR)
sqlalchemy	2.0.48	SQL toolkit and ORM
stack-data	0.6.3	Extract data from stack frames for tracebacks
starlette	1.0.0	Lightweight ASGI framework/toolkit
statsmodels	0.14.6	Statistical models, hypothesis tests and data exploration
strictyaml	1.7.3	Type-safe, schema-validated YAML parser
svgwrite	1.4.3	SVG file generation
swiglpk	5.0.13	SWIG bindings for GLPK linear optimizer
sympy	1.14.0	Symbolic mathematics
tblib	3.2.2	Pickle tracebacks and re-raise across processes
termcolor	3.3.0	ANSI color output in terminal
texttable	1.7.0	ASCII table formatting
texture2ddecoder	1.0.6	Decode GPU compressed texture formats
threadpoolctl	3.6.0	Control thread pool size for native libraries
tiktoken	0.12.0	Byte-pair encoding tokenizer (OpenAI)
tomli	2.4.1	TOML parser (read-only)
tomli-w	1.2.0	TOML file writer
toolz	1.1.0	Functional utilities — curry, pipe, compose
tqdm	4.67.3	Fast, extensible progress bar
traitlets	5.14.3	Type-checked attributes and configuration system
traits	7.1.0	Observable typed attributes
tree-sitter	0.23.2	Incremental parser for source code
tree-sitter-go	0.23.3	Go language grammar for tree-sitter
tree-sitter-java	0.23.4	Java language grammar for tree-sitter
tree-sitter-python	0.23.4	Python language grammar for tree-sitter
tskit	1.0.2	Tree sequence data structures for population genetics
typing-extensions	4.15.0	Backport of new typing module features
typing-inspection	0.4.2	Runtime inspection of type annotations
tzdata	2025.3	IANA timezone database as Python package
ujson	5.12.0	Ultra-fast JSON encoder/decoder
uncertainties	3.2.3	Transparent uncertainty propagation (error bars)
unyt	3.1.0	Unit handling and dimensional analysis
urllib3	2.6.3	HTTP client with connection pooling
vega-datasets	0.9.0	Sample datasets for Vega/Altair visualization
vrplib	2.1.0	Vehicle routing problem benchmark instances
wcwidth	0.6.0	Terminal character width (CJK and wide characters)
webencodings	0.5.1	Legacy web character encodings
wordcloud	1.9.6	Word cloud image generation
wrapt	2.1.2	Module for decorators, wrappers and monkey patching
xarray	2026.2.0	N-dimensional labeled arrays and datasets
xgboost	2.1.4	Gradient boosting framework (XGBoost)
xlrd	2.0.2	Read Excel .xls files
xxhash	3.6.0	xxHash non-cryptographic hash
xyzservices	2026.3.0	Tile map service providers for visualization
yarl	1.23.0	URL parsing and manipulation
yt	4.4.0	Volumetric data analysis and visualization
zarr	3.2.1	Chunked, compressed, N-dimensional arrays
zengl	2.7.2	Minimal OpenGL rendering context
zfpy	1.0.1	ZFP floating-point array compression
zstandard	0.25.0	Zstandard (zstd) compression bindings