pqready

Post-quantum cryptography readiness scanner. Audits TLS endpoints, source code, and certificate files for quantum-vulnerable crypto and surfaces a prioritised migration plan — as a CLI, a library, an MCP server, or a remote Cloudflare-hosted tool.

Install

pip install pqready

Requires Python 3.11+. Installs two console scripts:

Quickstart

# Scan a TLS endpoint
pqready tls api.example.com

# Scan a source tree
pqready source ./my-service

# Inspect a certificate file
pqready cert /etc/ssl/certs/api.example.com.pem

# Auto-detect target type
pqready scan ./my-service
pqready scan api.example.com

# Generate an HTML report alongside the terminal summary
pqready tls api.example.com --output report.html

Exit codes: 0 clean · 1 CRITICAL/HIGH findings · 2 scan error. Plays cleanly with CI: just call pqready and let the exit code fail the job.

What it detects

pqready looks for three classes of quantum-relevant weakness.

TLS endpoints — backed by sslyze

Source code — AST for Python, regex for config

Skips binary files and files > 1 MB. Prunes .git, .venv, node_modules, __pycache__, dist, build and friends.

Certificate files — PEM / DER / CRT / CER / KEY

Reuses the cert-level checks above (TLS-004 … TLS-009) and applies them to files on disk. Useful for offline audits of leaf certs and private keys.

Sample terminal output

pqready · TLS scan · api.example.com:443
────────────────────────────────────────
✗ CRITICAL  TLS-002  RSA key exchange offered (no perfect forward secrecy)
✗ CRITICAL  TLS-004  RSA certificate with 2048-bit key
✗ HIGH      TLS-005  ECC certificate (secp256r1)
✓ INFO      TLS-007  No hybrid PQC key exchange offered
Risk score: 80/100  ·  PQC-ready: NO
────────────────────────────────────────
Run with --output report.html to generate a full report.

The HTML report is self-contained (no external CSS / JS, no remote fetches). Open it in a browser or attach it to a ticket as-is.

Library usage

from pqready.core.tls import scan_tls
from pqready.core.source import scan_source
from pqready.core.certs import scan_cert_file
from pqready.reporters.html import render_html

tls_result   = scan_tls("api.example.com")
src_results  = scan_source("./services")
cert_result  = scan_cert_file("./api.example.com.pem")

render_html([tls_result, *src_results, cert_result], "report.html")

Every scanner returns a ScanResult:

@dataclass
class Finding:
    id: str               # e.g. "TLS-002"
    category: FindingCategory
    severity: Severity    # critical | high | medium | low | info
    title: str
    description: str
    location: str         # endpoint URL, file path, or "file:line"
    evidence: str
    remediation: str
    nist_ref: str         # e.g. "FIPS 203 (ML-KEM)"
    pqc_ready: bool

@dataclass
class ScanResult:
    target: str
    scan_type: str        # "tls" | "source" | "cert"
    findings: list[Finding]
    scanned_at: str       # ISO 8601
    duration_ms: int
    error: str | None
    risk_score: int       # 0-100, severity-weighted
    pqc_ready: bool       # False if any CRITICAL/HIGH

MCP server

pqready ships an MCP server (pqready-mcp) over stdio that exposes four tools any MCP-capable agent (Claude Code, Cursor, Cline, …) can call:

Wire it in by adding to your MCP client config:

{
  "mcpServers": {
    "pqready": { "command": "pqready-mcp" }
  }
}

Full tool schemas, example payloads, and a decision tree for which tool to call live in skill/SKILL.md — drop that file into any agent skill loader to teach the agent how to use pqready correctly.

Cloudflare Worker (remote MCP)

A thin proxy Worker lives in cf_worker/. It handles bearer auth and streams MCP traffic to a backend running pqready-mcp over HTTP/SSE (Cloud Run, Fly, VPS, etc.). The Python runtime cannot execute directly in a Worker, so this layer is auth + proxy only.

cd cf_worker
npm install
wrangler secret put PQREADY_API_TOKEN
npm run deploy

See cf_worker/README.md for the full setup.

Interpreting findings

risk_score is a coarse summary, not a substitute for reading the findings. Two endpoints can hit 100 for entirely different reasons.

pqc_ready is intentionally conservative: it returns False whenever any CRITICAL or HIGH finding exists. A green run only means we did not find any of the patterns we know to look for — it is not a certification.

Every finding carries:

• NIST reference — most often FIPS 203 (ML-KEM), FIPS 204 (ML-DSA), SP 800-52, or SP 800-131A.
• Remediation — concrete, copy-pasteable next step.

See skill/SKILL.md for the full per-ID remediation table.

Documentation

Development

git clone https://github.com/pqready/pqready.git
cd pqready
python -m venv .venv && source .venv/bin/activate   # or .venv\Scripts\activate on Windows
pip install -e ".[dev]"
ruff check .
pytest tests/

CI runs ruff + pytest on Python 3.11 and 3.12. Releases are cut by pushing a v* tag; .github/workflows/publish.yml publishes to PyPI via OIDC trusted publishing — no API tokens stored in the repo.

Contributing

Bug reports and PRs welcome. For non-trivial changes, please open an issue first so we can align on scope. Run ruff check . and pytest tests/ before opening a PR.

Limitations

• TLS scans depend on the handshake sslyze observes. Some load balancers hide cipher details from external probes.
• The source scanner is pattern-based and will miss findings hidden behind dynamic dispatch (e.g. getattr(crypto_mod, name)).
• PQC group detection matches OpenSSL-style names. Servers exposing only IANA codepoints in custom log formats may report differently.
• pqready does not test certificate revocation or CT-log inclusion.

License

Apache-2.0 — see LICENSE.