Plexavo

An open-source AWS misconfiguration scanner that reads your account with your own local AWS credentials — nothing is ever handed to anyone else. Run it yourself, the same way you'd run aws s3 ls, and get a 0-100 security score plus a plain-English report explaining exactly what's wrong, what an attacker would actually do with it, and the exact command to fix it.

Deterministic detection (pure Python/boto3, never AI) finds the misconfigurations. Claude only rewrites already-computed technical findings into a narrative a non-security founder can act on — it never decides what counts as a finding, and it's entirely optional (see Cost below).

What it checks

31 checks across 6 categories, all validated against real, live AWS accounts (not just offline logic) unless a specific, stated limitation made that impossible — see the individual docs/*-TEST-MATRIX.md files for exactly which checks are fully verified and which have a documented, structural reason they can't be (e.g. some checks can only prove their "clean" case without disabling a real security control to test the "bad" case).

Plus scoring.py (the 0-100 score) and plexavo/report/ai_narration.py (the opt-in AI layer).

Example output

Illustrative example (a public S3 bucket, an unencrypted volume, a role that's never been assumed) — this is what a scan actually surfaces, including a finding's free template remediation (shown by default, no --explain needed; --explain would replace this panel with a full AI narrative instead):

Your AWS Security Score: 74/100 (Good)

                                              Findings (3)
┏━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Check   ┃ Severity ┃ Resource                           ┃ Detail                                     ┃
┡━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
│ STOR-19 │ Critical │ arn:aws:s3:::taskflow-uploads-prod │ Bucket 'taskflow-uploads-prod' allows      │
│         │          │                                    │ public read via bucket policy.             │
│ ENC-29  │ Medium   │ vol-015506d04acb46ca9              │ EBS volume 'vol-015506d04acb46ca9' is not  │
│         │          │                                    │ encrypted at rest.                         │
│ USE-27  │ High     │ AWSSecurityScannerReadOnlyRole     │ Role 'AWSSecurityScannerReadOnlyRole' has  │
│         │          │                                    │ never been assumed since creation.         │
└─────────┴──────────┴────────────────────────────────────┴────────────────────────────────────────────┘

╭────────────────────────────────────────── source: template ──────────────────────────────────────────╮
│ STOR-19 — arn:aws:s3:::taskflow-uploads-prod                                                         │
│                                                                                                      │
│ IMPACT: Bucket 'taskflow-uploads-prod' doesn't have full S3 Block Public Access protection enabled.  │
│ Without this protection, a single mistake — an overly broad bucket policy, a public ACL grant,       │
│ someone copy-pasting a policy from a tutorial — immediately exposes every object to anyone on the    │
│ internet via a plain s3:GetObject call, with nothing left to catch the mistake.                      │
│                                                                                                      │
│ CONFIDENCE: Confirmed                                                                                │
│                                                                                                      │
│ NEXT STEP: Turn on Block Public Access now: aws s3api put-public-access-block --bucket               │
│ taskflow-uploads-prod --public-access-block-configuration                                            │
│ "BlockPublicAcls=true,IgnorePublicAcls=true,BlockPublicPolicy=true,RestrictPublicBuckets=true"       │
│                                                                                                      │
│ FULL FIX DETAIL: Enable all four Block Public Access settings unless there's a specific, documented  │
│ reason not to:                                                                                       │
│ aws s3api put-public-access-block --bucket taskflow-uploads-prod --public-access-block-configuration │
│ "BlockPublicAcls=true,IgnorePublicAcls=true,BlockPublicPolicy=true,RestrictPublicBuckets=true"        │
╰──────────────────────────────────────────────────────────────────────────────────────────────────────╯

Severity, confidence, and evidence are always separate signals, never merged into one label — a low-confidence Critical and a high-confidence Medium don't read the same. This particular finding has no Evidence line because it's a directly-observed fact with nothing uncertain about it — Evidence only appears when a check has a concrete account-state fact behind it (e.g. an IAM policy scoped by a Condition block, or a CloudTrail lookup that hit its page cap on USE-26), and Confidence only drops from "Confirmed" in that same situation. --report-html/--report-pdf render this same data as a full report; see Cost for what --explain needs and what it costs.

Install

Plexavo is a command-line tool. Pick your OS below — every path installs it into its own isolated environment, so it never clashes with your other Python packages.

macOS & Linux

uv puts a single plexavo command on your PATH, in every terminal, with nothing to activate:

curl -LsSf https://astral.sh/uv/install.sh | sh   # skip if you have uv
uv tool install plexavo

Update or remove later with uv tool upgrade plexavo / uv tool uninstall plexavo. Prefer pipx? pipx install plexavo works the same way.

Windows

uv and pipx install fine, but the small plexavo.exe launcher they drop on your PATH is unsigned, and Windows Smart App Control refuses to run unsigned executables it doesn't recognise. The way around it is to run Plexavo through Python directly. Two ways — both isolated, pick one:

Option 1 — uv (recommended)

uv tool install plexavo

Set-ExecutionPolicy -Scope CurrentUser RemoteSigned

if (!(Test-Path $PROFILE)) { New-Item -ItemType File -Path $PROFILE -Force }
Add-Content $PROFILE 'function plexavo { & "$env:APPDATA\uv\tools\plexavo\Scripts\python.exe" -m plexavo @args }'

Open a new terminal and plexavo works exactly like it does on macOS/Linux.

uv tool install still leaves the blocked plexavo.exe on your PATH. The line you add to your PowerShell profile shadows it with a command that calls uv's bundled python.exe directly — that Python is signed by the Python Software Foundation, so Smart App Control never stops it. You add it once; the Set-ExecutionPolicy line (also once) is what lets PowerShell load your profile at all. Update later with uv tool upgrade plexavo.

Option 2 — virtual environment (no uv)

py -m venv plexavo-venv
.\plexavo-venv\Scripts\Activate.ps1
python -m pip install plexavo
python -m plexavo

Use python -m for everything inside the venv. Creating a venv also generates its own pip.exe and plexavo.exe, and those are exactly the unsigned launcher stubs Smart App Control blocks — but the venv's python.exe is a copy of your real Python with its signature intact, so going through it (python -m pip, python -m plexavo) always works.

Activating the venv (.\plexavo-venv\Scripts\Activate.ps1) just points python at this environment for the current terminal — you re-run it in each new terminal, and deactivate leaves it. If PowerShell blocks the activate script, run Set-ExecutionPolicy -Scope CurrentUser RemoteSigned once. Update with python -m pip install --upgrade plexavo inside the activated environment.

AI-narrated explanations (optional, any OS)

Plexavo is complete without this. If you have an Anthropic API key and want each finding rewritten as a full narrative (see Cost), install "plexavo[ai]" in place of plexavo in any command above.

From source (for contributing, or trying an unreleased change)

git clone https://github.com/plexavo/plexavo.git
cd plexavo
uv pip install -e .   # or: pip install -e . (inside a venv)

Using Plexavo

Run it with no arguments and it walks you through everything — choosing an AWS profile (or setting a new one up), picking whether you want an HTML or PDF report, then running the scan and showing your 0-100 score with every finding and its plain-English fix. Nothing to memorise.

plexavo             # macOS/Linux, and Windows Option 1
python -m plexavo   # Windows Option 2 — inside the activated venv

Scripting a scan into CI or a scheduled job? Add scan and flags — plexavo scan --help (or python -m plexavo scan --help) covers the flag-driven form (--profile, --region, --report-html, --report-pdf, --explain).

Project structure

plexavo/
├── auth.py                    # local AWS credential resolution
├── principals.py               # enumerates IAM users/roles + their policies
├── findings.py                  # Finding data model, Severity enum
├── scoring.py                    # 0-100 score from a list of Findings
├── cli.py                         # `plexavo scan ...` entry point
├── __main__.py                      # lets `python -m plexavo` run the CLI
├── checks/
│   ├── iam.py                       # IAM-01 to IAM-06 (privilege escalation)
│   ├── iam_hygiene.py                # IAM-07 to IAM-14 (hygiene, cross-account trust)
│   ├── network.py                     # NET-01 to NET-04
│   ├── storage.py                      # STOR-19 to STOR-21
│   ├── encryption.py                    # ENC-29 to ENC-31
│   ├── logging.py                        # LOG-22 to LOG-25
│   └── usage.py                           # USE-26, USE-27
└── report/
    ├── html_report.py               # assembles findings into HTML via Jinja2
    ├── pdf.py                        # same data, rendered to PDF via fpdf2
    ├── ai_narration.py                # opt-in Claude narration + the 10 free templates
    ├── fonts/                          # bundled DejaVu Sans (PDF) + Geist (HTML) —
    │                                     both self-hosted, zero external requests
    └── templates/report.html.j2         # the HTML report template

examples/
└── quickstart-sandbox.tf       # one cheap, deliberately-public S3 bucket —
                                  try the scanner without pointing it at
                                  real infrastructure on day one

tests/
└── test_*.py                   # one per module, no AWS calls, run anytime:
                                  python tests/test_iam_offline.py, etc.

docs/
├── *-TEST-MATRIX.md            # one per category — the actual grading
│                                  record: what's verified, how, and any
│                                  stated limitation. Start here if you
│                                  want to know how much to trust a
│                                  given check.
├── COMPARISON.md               # honest, hands-on comparison against
│                                  Prowler, ScoutSuite, and PMapper
├── TECHNICAL-EXPLAINER.md      # implementation decisions and why
└── HOW-IT-WORKS.md

Cost

Detection is free (pure Python/boto3), always, regardless of anything else in this section. Every scan also gets free Next Step / Full Fix Detail remediation wherever one of 10 hand-written templates matches the finding type — no flag, no API key, zero cost, on by default.

Live AI only runs with --explain, and when it does it's used for every finding, including the 10 templated ones (a deliberate choice — "AI narration on" always means fully AI-written content, not a mix of template and AI), typically $0.01-0.02 per finding depending on answer length. A full scan with --explain on a real account is usually well under a dollar. This is your own ANTHROPIC_API_KEY, in your own Anthropic account — this project never sees your key, never embeds one of its own, and never calls the API on your behalf without you having set one. Not "free AI" — bring your own key, and a scan with it enabled costs a few cents.

Running the tests

pip install -e ".[dev]"
for f in tests/test_*.py; do python "$f"; done

Each is self-contained — fake AWS API responses, no real credentials or network calls needed. See docs/TEST-MATRIX.md for how these map to live-AWS verification.

Contributing

See CONTRIBUTING.md — in particular, the pattern for adding a new check.

Security

Found a vulnerability in the tool itself (not a misconfiguration in your own AWS account — that's the tool working correctly)? See SECURITY.md for a private reporting path.

License

AGPL-3.0 — see LICENSE. You can use, run, and modify this freely. If you run a modified version as a hosted service, you're required to publish those modifications too. This is deliberate: it's the specific protection against a well-resourced company taking this code and standing up a competing hosted product without ever contributing back.