Skip to main content

SIOS — Financial Audit Engine

Detect hidden financial losses in your transaction data. Every finding is reproducible and cryptographically signed.

pip install sios
sios run examples/sample.csv
────────────────────────────────────────────────────────
  SIOS Audit — sample.csv
  21 transactions analyzed
────────────────────────────────────────────────────────

  [Duplicate payment]        1,195 EUR  conf: 95%
  Duplicate AWS charge — EC2 Production (Feb)

  [Cost anomaly]             7,650 EUR  conf: 72%
  AWS spike vs. baseline — load test environment (Feb 14)

  [Unused subscription]        588 EUR  conf: 80%
  GitHub Teams — 3 recurring charges, no usage signal

  [Cloud waste]                522 EUR  conf: 65%
  EC2 dev environments running 24/7 across 3 months

────────────────────────────────────────────────────────
  Estimated recoverable: 31,121 EUR
  Findings: 12
────────────────────────────────────────────────────────

One dataset → one audit → one proof.


Install

pip install sios

Requires Python 3.11+. Zero configuration.


Quickstart

# Run a full financial audit
sios run transactions.csv

# JSON output for downstream processing
sios detect transactions.csv --format json

# Export to file
sios export transactions.csv --out report.json

Time-to-value: under 2 minutes from install to first findings.


What SIOS detects

Finding type Description Typical recovery
duplicate_payment Same vendor + similar amount within 7 days Full refund
unused_subscription Recurring charges with no counterpart activity Cancel or renegotiate
cost_anomaly Statistically abnormal spend (IQR fence per vendor) Investigate + credit
cloud_waste Dev/staging environments running continuously Right-size or terminate

Each finding includes an estimated recovery amount, a confidence score (0–100%), an evidence snapshot, and recommended actions.


Input format

CSV with these columns (order doesn't matter):

date,amount,currency,vendor,description
2024-01-15,299,EUR,Slack,Slack Pro subscription
2024-02-15,299,EUR,Slack,Slack Pro subscription
2024-01-20,1250,EUR,AWS,EC2 dev server

Minimum required: date, amount, vendor.
Also accepts JSON. Currency defaults to EUR; detected per-dataset automatically.


Python SDK

from sios import SIOS

agent = SIOS()
result = agent.run("data/transactions.csv")

print(f"Estimated savings: {result.estimated_savings:,.0f} {result.currency}")

for finding in result.findings:
    print(f"  [{finding.type.value}]  {finding.estimated_amount:,.0f} {finding.currency}  conf={finding.confidence:.0%}")
    print(f"  {finding.title}")
    print(f"  {finding.description[:100]}")

AuditResult fields

Field Type Description
findings List[Finding] Detected anomalies, ordered by amount
estimated_savings float Total recoverable across all findings
currency str Currency code (EUR, USD, …)
dataset_rows int Number of transactions analyzed
summary dict Breakdown by finding type

Cloud connectors

AWS Cost Explorer

Pull your AWS billing data directly — no CSV export needed:

pip install sios[aws]

sios aws --days 90
sios aws --profile production --days 60 --save aws_costs.csv

Requires IAM permission ce:GetCostAndUsage. Detects idle reservations, dev environment waste, and service-level anomalies.

Stripe

Pull charges and subscriptions, detect duplicate billing and subscription drift:

pip install sios[stripe]

export STRIPE_API_KEY=sk_live_...
sios stripe --days 90
sios stripe --api-key sk_live_... --save stripe_charges.csv

Python API — connectors

from sios.connectors.aws import AWSConnector
from sios.connectors.stripe import StripeConnector
from sios.value_engine.engine import ValueEngine

# AWS
transactions = AWSConnector(profile="production").fetch(days=90)
findings = ValueEngine().run(transactions)

# Stripe
transactions = StripeConnector(api_key="sk_live_...").fetch(days=90)
findings = ValueEngine().run(transactions)

Verifiable proofs (optional)

Every finding can be signed and submitted to a Proof Protocol node:

sios prove transactions.csv --node https://your-sios-node.onrender.com
Generating proofs via https://your-sios-node.onrender.com ...

  AWS duplicate charge confirmed          CPO: 3f8a21c9b7...
  Slack subscription anomaly              CPO: 9d1e4507a2...
  Dev environment cloud waste             CPO: c2f6bb0814...

3 proofs generated.

Each CPO (Computational Proof Object) is:

  • Cryptographically signed — Ed25519 signature over the canonical payload
  • Content-addressed — SHA-256 hash of inputs + outputs
  • Reproducible — re-execute at any time, get the same result
  • Append-only — recorded in a tamper-evident ledger

The proof ties the finding to the specific code, data, and environment that produced it — creating an audit trail suitable for sharing with finance teams, external auditors, or vendors during dispute resolution.


Export

# JSON report
sios detect transactions.csv --format json > report.json

# CSV for Excel / BI tools
sios detect transactions.csv --format csv > report.csv

# Save directly
sios export transactions.csv --out report.json
sios export transactions.csv --format csv --out report.csv

Architecture

  Input (CSV · JSON · Stripe · AWS)
          │
          ▼
  ┌─────────────────────┐
  │  Ingestion layer    │  Transaction normalization → canonical data model
  └─────────────────────┘
          │
          ▼
  ┌─────────────────────┐
  │  Value Engine       │  Rule-based detection pipeline (idempotent, stateless)
  │  4 detectors        │  duplicate · subscription · anomaly · cloud waste
  └─────────────────────┘
          │
          ▼
  ┌─────────────────────┐
  │  Findings           │  amount · confidence · evidence · recommended actions
  └─────────────────────┘
          │
          ▼
  ┌─────────────────────┐
  │  Proof Protocol     │  Ed25519 sign · SHA-256 hash · append-only ledger
  │  (optional)         │  Verifiable, reproducible execution trace
  └─────────────────────┘

Design principles:

  • Deterministic execution — same input always produces the same findings
  • Idempotent processing — running twice on the same dataset returns cached results
  • Modular architecture — each detector is an independent plugin; add your own
  • Persistent store — SQLite (default) or Postgres; WAL mode, thread-safe
  • Stateless API — the FastAPI server layer adds no shared mutable state

CLI reference

sios run    file.csv                      # Full audit, formatted output
sios detect file.csv                      # Detection only, table format
sios detect file.csv --format json        # JSON output
sios detect file.csv --format csv         # CSV output
sios prove  file.csv --node <url>         # Audit + generate CPO proofs
sios export file.csv --out report.json    # Save to file
sios aws    --days 90                     # Pull from AWS Cost Explorer
sios stripe --days 90                     # Pull from Stripe

Install options

pip install sios              # Core (CLI + Python SDK)
pip install sios[aws]         # + AWS Cost Explorer connector
pip install sios[stripe]      # + Stripe connector
pip install sios[server]      # + FastAPI proof server
pip install sios[all]         # Everything

Why SIOS

Most companies lose money silently:

  • Forgotten SaaS subscriptions — tools nobody uses, renewed automatically
  • Duplicate vendor billing — the same invoice paid twice in the same cycle
  • Cloud environments running 24/7 — dev and staging never turned off
  • One-time spikes never investigated — a load test billed at full rate, forever forgotten

SIOS detects these patterns automatically from raw transaction data.
Every result is backed by a confidence score, an evidence snapshot, and — when you need it — a cryptographic proof that the finding is reproducible and tamper-evident.

Install, run, recover value.


Use cases

User How they use SIOS
Finance team Monthly audit of card statements and cloud bills
Engineering Post-incident cost review after infrastructure changes
Startup CFO SaaS spend audit before a funding round
Accountant / firm Client cost optimization as a service
SaaS vendor Embed duplicate billing detection into your own product

Roadmap

  • CLI (sios run, detect, prove, export)
  • Python SDK (from sios import SIOS)
  • AWS Cost Explorer connector
  • Stripe connector
  • Verifiable proofs (CPO)
  • PyPI package (pip install sios)
  • PDF / bank statement ingestion
  • Qonto, Pennylane, QuickBooks, Sage connectors
  • Continuous monitoring (scheduled detection)
  • Export to PDF audit report
  • Custom detector plugin API

License

MIT — free to use, modify, and embed in commercial products.


SIOS detects financial inefficiencies in structured data. We turn raw transactions into audited savings opportunities.

Metadata

Release files for sios 1.0.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for sios 1.0.0
File Size Uploaded Provenance Actions
sios-1.0.0.tar.gz 465.9 kB With provenance Details for sios-1.0.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for sios 1.0.0
File Interpreter ABI Platform Actions
sios-1.0.0-py3-none-any.whl Python 3 none any Details for sios-1.0.0-py3-none-any.whl

Total release size: 503.9 kB

Release files / sios-1.0.0.tar.gz

Download URL sios-1.0.0.tar.gz
Size 465.9 kB
Tags Source
SHA-256 checksum
How to use checksums
99fcad051841b7d1070f172b63689ba8415f3b115728ed8949ce9b9cb3bc4d33
BLAKE2b-256 checksum
How to use checksums
bed35dd8e827f77ada77d8db5fea43b3d2a65cec0c89f75d045ae44657bf7a49
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 28, 2026.

Transparency log

Release files / sios-1.0.0-py3-none-any.whl

Download URL sios-1.0.0-py3-none-any.whl
Size 38.0 kB
Tags Python 3
SHA-256 checksum
How to use checksums
9ce5c2089c8e8288ccab961a96c2d532e5fc284996e9e1f4f56d2002c8055f41
BLAKE2b-256 checksum
How to use checksums
f286044d594320924c8beda1c900ed94bb34de3e25fd91131dd0758f99b169c1
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 28, 2026.

Transparency log

Security for sios 1.0.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

Both files in this release have been uploaded with provenance metadata.

Select from 2 files

Attestation for sios-1.0.0.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 28, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

1.0.0 This release

2 release files Full provenance

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