silueta

Profile the shape of your data — never the values.

silueta turns a dataset — CSV, Parquet, and soon multi-tab Excel and databases — into a compact, machine-readable profile of its shape: types, nulls, cardinality, value-shape masks, semantic PII/PHI flags, and red-flag anomalies. The profile contains no raw cell values, by construction, so an AI agent (or a teammate, or a ticket) can work from it safely.

The motivating workflow: you get a vendor workbook full of PII/PHI and need to model it — dbt sources, Snowflake landing tables, relationships. Most of what modeling needs is how the data behaves, not what it says. silueta captures the behavior so an agent like Claude Code can scaffold the models for data it is never allowed to read.

$ uvx silueta scan patients.csv --report report.md
wrote profile.json
wrote report.md
  patients: 30 rows, 7 columns, 2 alerts

The profile tells the agent things like:

  • mrn — VARCHAR, 100% unique, mask AAA-999999 → this is your key
  • ssn — matches SSN structure at 100% with plausibility checks → sensitive, alert
  • amount — typed as text but 100% castable numeric, magnitude 10^2 → DECIMAL, fix at landing
  • clinic — constant column (and that is all it says about it)
  • visit_date — dates spanning 2025 (year granularity only)

The contract

A silueta profile never contains:

  • raw cell values, sample rows, or frequent-value lists
  • min/max for strings or dates (those are raw values) — numerics report order-of-magnitude and precision facts, dates report year spans
  • serialized sketches or value hashes (hashing is not anonymization)
  • any statistic derived from values observed in fewer than k rows (default 5); constant columns report nothing but their constancy

This contract is enforced by tests that scan every output for source values, and it travels inside the artifact itself (profile.json → silueta.contract).

What it is and is not: silueta protects against accidental context leakage — values ending up in an LLM conversation, a log, a pasted ticket. It does not claim formal anonymity (differential privacy) against an adversary reconstructing small populations from aggregates. The wording is "no raw values by construction," never "anonymized."

Status

Early alpha — CSV and Parquet profiling work today. On deck (see issues): messy-Excel header/type recovery (the headline feature), cross-table foreign-key candidates via exact containment, dbt/Snowflake scaffold emitters, database connections.

Install

uvx silueta            # zero-install run
uv tool install silueta
pipx install silueta

License

MIT

Project links

Data verified by PyPI on Jul 24, 2026
Data provided by the project maintainers, verified at the time the release was uploaded to PyPI.

Key dates

PyPI data
Data sourced directly from PyPI's database.
  • Released:
Newer release available (0.3.0)

1 maintainer

PyPI data
Data sourced directly from PyPI's database.
Avatar for chekos from gravatar.com chekos

Credits

Author: Sergio Sánchez Zavala

GitHub Statistics

Data verified by PyPI on Jul 24, 2026
The GitHub source repository was provided by the project maintainers and verified by PyPI at the time of upload. Stars, forks, and open issues/PRs are derived from that repository and have not been independently verified.

License expression

MIT
View SPDX License List

Requires

Python >=3.10

Tags

agents data-profiling dbt llm phi pii privacy snowflake

Classifiers

Report project as malware

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

silueta-0.0.1.tar.gz (37.1 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

silueta-0.0.1-py3-none-any.whl (12.6 kB view details)

Uploaded Python 3

File details

Details for the file silueta-0.0.1.tar.gz.

File metadata

  • Download URL: silueta-0.0.1.tar.gz
  • Upload date:
  • Size: 37.1 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.14

File hashes

Hashes for silueta-0.0.1.tar.gz
Algorithm Hash digest
SHA256 c83de817e85100142ae6ec214a3ca15f791426055e6e9dead4819de67b20f211
MD5 912ac3b506589ca7153cca7a4861fccb
BLAKE2b-256 67771f6d2c0e4107b178c57fa3601d557046ccf55a239239ac99698ef56c0758

See more details on using hashes here.

Provenance

The following attestation bundles were made for silueta-0.0.1.tar.gz:

Publisher: release.yml on chekos/silueta

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file silueta-0.0.1-py3-none-any.whl.

File metadata

  • Download URL: silueta-0.0.1-py3-none-any.whl
  • Upload date:
  • Size: 12.6 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.14

File hashes

Hashes for silueta-0.0.1-py3-none-any.whl
Algorithm Hash digest
SHA256 7438f10af8260ff1d63f0475281b9deff6ef3c2a867a1e7b8bd26d8028e8e723
MD5 73c669483f76112b43eeb36c4eca6a0e
BLAKE2b-256 bbbd5a195686df3df7cde311635625ea5ffcd84ee194de78985d846f2b2d447a

See more details on using hashes here.

Provenance

The following attestation bundles were made for silueta-0.0.1-py3-none-any.whl:

Publisher: release.yml on chekos/silueta

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Release history Release notifications | RSS feed

0.3.0

2 files

0.2.0

2 files

0.1.0

2 files

This release

0.0.1 This release

2 files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page