Fraudlens

pip install cognis-fraudlens
fraudlens scan .            # → prioritized findings in seconds

🔎 Example output

Real, reproducible output from the tool — runs offline:

$ fraudlens-emit --version
fraudlens 0.1.0

$ fraudlens-emit --help
usage: fraudlens [-h] [--version] {backtest,rules} ...

Replay transactions against pluggable fraud rules and report precision/recall, alert volume, and a caught-vs-missed diff.

positional arguments:
  {backtest,rules}
    backtest        replay a labeled transaction CSV against the ruleset
    rules           list available fraud rules

options:
  -h, --help        show this help message and exit
  --version         show program's version number and exit

Command-line interface for FRAUDLENS.

Examples
--------
  # Backtest the default ruleset against a labeled CSV (human-readable table)
  python -m fraudlens backtest transactions.csv

  # JSON output for CI / piping into jq
  python -m fraudlens backtest transactions.csv --format json | jq .metrics

  # Only run a subset of rules
  python -m fraudlens backtest transactions.csv --rules high_amount,velocity

  # Override a threshold and fail CI if recall drops below 0.8
  python -m fraudlens backtest transactions.csv \
      --set high_amount_threshold=500 --min-recall 0.8

  # List available rules
  python -m fraudlens rules

Exit codes
----------
  0  success and all gates (if any) passed
  1  a quality gate (--min-recall / --min-precision / --max-alert-rate) failed
  2  bad usage / unparseable input

$ fraudlens-emit rules
Available fraud rules:
  high_amount      transaction amount over threshold
  velocity         rapid-fire transactions on one account
  odd_hour         sizeable spend during overnight hours
  foreign_geo      spend outside home country

Blocks above are real fraudlens output — reproduce them from a clone.

Usage — step by step

1. Install the CLI:

   pipx install "git+https://github.com/cognis-digital/fraudlens.git"
   

2. List the available fraud rules, then backtest a labeled transaction CSV against them (primary command):

   fraudlens rules
   fraudlens backtest transactions.csv
   

3. Iterate on the ruleset — enable a subset and override thresholds without editing code:

   fraudlens backtest transactions.csv \
     --rules high_amount,velocity \
     --set high_amount_threshold=500
   

4. Read the output — precision / recall / alert rate as a table, or JSON for diffing. Set gates so a regression exits non-zero:

   fraudlens backtest transactions.csv --format json > metrics.json
   fraudlens backtest transactions.csv --min-recall 0.8 --max-alert-rate 0.05
   

5. Automate in CI — fail the job when a rule change drops recall or floods alerts:

   fraudlens backtest data/labeled.csv --min-recall 0.85 --min-precision 0.6
   # non-zero exit => the rule change regressed
   

Contents

• Why fraudlens? · Features · Quick start · Example · Architecture · AI stack · How it compares · Integrations · Install anywhere · Related · Contributing

Why fraudlens?

Fraud teams iterate rules in notebooks with no reproducible CLI harness. Backtest a rule change against historical CSV/Parquet and get a diff of caught-vs-missed fraud as a CI artifact.

fraudlens is single-purpose, scriptable, and self-hostable: point it at a target, get prioritized results in the format your workflow already speaks (table · JSON · SARIF), gate CI on it, and let agents drive it over MCP.

Features

• ✅ Parse Transactions
• ✅ Load Transactions
• ✅ Build Ruleset
• ✅ List Rules
• ✅ Backtest
• ✅ Report To Dict
• ✅ To Json
• ✅ Runs on Linux/macOS/Windows · Docker · devcontainer
• ✅ Ports in Python, JavaScript, Go, and Rust (ports/)

Quick start

pip install cognis-fraudlens
fraudlens --version
fraudlens scan .                       # scan current project
fraudlens scan . --format json         # machine-readable
fraudlens scan . --fail-on high        # CI gate (non-zero exit)

Example

$ fraudlens scan .
  [HIGH    ] FRA-001  example finding             (./src/app.py)
  [MEDIUM  ] FRA-002  another signal              (./config.yaml)

  2 findings · risk score 5 · 38ms

Architecture

flowchart LR
  IN[input] --> P[fraudlens<br/>analyze + score]
  P --> OUT[report]

Use it from any AI stack

fraudlens is interoperable with every popular way of using AI:

• MCP server — fraudlens mcp (Claude Desktop, Cursor, Cognis.Studio, uncensored-fleet)
• OpenAI-compatible / JSON — pipe fraudlens scan . --format json into any agent or LLM
• LangChain · CrewAI · AutoGen · LlamaIndex — wrap the CLI/JSON as a tool in one line
• CI / scripts — exit codes + SARIF for non-AI pipelines

How it compares

	Cognis fraudlens	Feedzai
Self-hostable, no account	✅	varies
Single command, zero config	✅	⚠️
JSON + SARIF for CI	✅	varies
MCP-native (AI agents)	✅	❌
Polyglot ports (JS/Go/Rust)	✅	❌
Open license	✅ COCL	varies

Built in the spirit of Feedzai / IEEE-CIS fraud kernels, re-framed the Cognis way. Missing a credit? Open a PR.

Integrations

Pipes into your stack: SARIF for code-scanning, JSON for anything, an MCP server (fraudlens mcp) for AI agents, and a webhook forwarder for SIEM/Slack/Jira. See docs/INTEGRATIONS.md.

Install — every way, every platform

pip install "git+https://github.com/cognis-digital/fraudlens.git"    # pip (works today)
pipx install "git+https://github.com/cognis-digital/fraudlens.git"   # isolated CLI
uv tool install "git+https://github.com/cognis-digital/fraudlens.git" # uv
pip install cognis-fraudlens                                          # PyPI (when published)
docker run --rm ghcr.io/cognis-digital/fraudlens:latest --help        # Docker
brew install cognis-digital/tap/fraudlens                             # Homebrew tap
curl -fsSL https://raw.githubusercontent.com/cognis-digital/fraudlens/main/install.sh | sh

Linux	macOS	Windows	Docker	Cloud
scripts/setup-linux.sh	scripts/setup-macos.sh	scripts/setup-windows.ps1	docker run ghcr.io/cognis-digital/fraudlens	DEPLOY.md (AWS/Azure/GCP/k8s)

Related Cognis tools

• panhound — Scans code, logs, fixtures, and S3 buckets for leaked PANs (Luhn-validated card numbers) and CVVs before they hit prod.
• obscan — Conformance and security linter for Open Banking / FAPI APIs: validates OAuth flows, consent scopes, and PSD2 endpoints against the spec.
• ledgerproof — Verifies double-entry ledger integrity and tamper-evidence by checking balance invariants and hash-chained journal entries.
• iso20022 — Validates, lints, and diffs ISO 20022 / pacs / camt payment messages and translates legacy MT into MX with schema-aware errors.
• tokenvault — Self-hostable PCI tokenization microservice and CLI that swaps PANs for format-preserving tokens and proves no raw card data persists.
• sanctscan — Screens counterparties and transactions against OFAC/EU/UN sanctions lists with fuzzy name matching and explainable hit scoring.

Explore the suite → 🗂️ all 170+ tools · ⭐ awesome-cognis · 🔗 cognis-sources · 🤖 uncensored-fleet · 🧠 engram

Contributing

PRs, new rules, and demo scenarios are welcome under the collaboration-pull model — see CONTRIBUTING.md and SECURITY.md.

⭐ If fraudlens saved you time, star it — it genuinely helps others find it.

Interoperability

{} composes with the 300+ tool Cognis suite — JSON in/out and a shared OpenAI-compatible /v1 backbone. See INTEROP.md for the suite map, composition patterns, and reference stacks.

License

Source-available under the Cognis Open Collaboration License (COCL) v1.0 — free for personal, internal-evaluation, research, and educational use; commercial / production use requires a license ([email protected]). See LICENSE.

---