nosecrets 0.2.0

Scan for hardcoded secrets before they reach your repository

nosecrets

Scan for hardcoded secrets before they reach your repository.

Zero runtime dependencies. Runs as a pre-commit hook or standalone CLI. Designed for developers who want a lightweight, auditable secrets scanner without pulling in a 20MB Go binary or a heavy Python dependency tree.

$ secrets scan

  [CRITICAL]  src/config.py:14
               AWS Access Key ID: AKIA************MPLE

  [HIGH]  deploy/docker-compose.yml:8
           Database Connection String: postg****alhost

  Found 2 potential secret(s): 1 CRITICAL, 1 HIGH

  To suppress a finding:
    Inline:  add  # nosecrets  at end of line
    Global:  add path to .secretsignore

---

Install

pip install nosecrets

Requires Python 3.12+.

---

Usage

# Scan current directory
secrets scan

# Scan a specific path
secrets scan src/

# Scan only staged files (pre-commit)
secrets scan --staged

# JSON output for CI
secrets scan --json

# Only report HIGH and above
secrets scan --severity high

# Install pre-commit hook
secrets init

# Remove hook
secrets uninstall

---

Pre-commit hook

secrets init

Installs a git hook that runs secrets scan --staged before every commit. If a CRITICAL or HIGH secret is found, the commit is blocked.

To remove:

secrets uninstall

---

What it detects

Severity	Pattern
CRITICAL	RSA / EC / OpenSSH private keys
CRITICAL	AWS Access Key ID (AKIA...)
CRITICAL	AWS Secret Access Key (with keyword context)
CRITICAL	Azure Storage connection string (DefaultEndpointsProtocol=...)
CRITICAL	Azure Storage Account Key (with keyword context)
CRITICAL	OpenAI API Key (sk-..., sk-proj-...)
CRITICAL	Anthropic API Key (sk-ant-...)
CRITICAL	Stripe Live Secret Key (sk_live_...)
HIGH	GitHub Tokens (ghp_, github_pat_, gho_, ghs_, ghr_)
HIGH	GitLab Personal Access Token (glpat-...)
HIGH	npm Access Token (npm_...)
HIGH	Google API Key (AIza...)
HIGH	Slack Bot / User / Webhook tokens
HIGH	SendGrid API Key (SG....)
HIGH	Twilio Account SID (AC...)
HIGH	Database connection strings with credentials
MEDIUM	JSON Web Tokens (hardcoded)
MEDIUM	Generic secrets (high-entropy api_key, password, token values)

---

False positive prevention

Two-layer defence against false positives:

1. Placeholder detection — values like "changeme", "your_api_key_here", ${SECRET}, {{token}} are never flagged regardless of pattern match.

2. Shannon entropy filter — generic patterns (password, api_key, token) only trigger when the value has sufficient entropy. password = "test123" is ignored. password = "X7#mK9$pL2@nQ8!" is flagged.

High-confidence patterns (AWS AKIA..., GitHub ghp_..., etc.) skip entropy filtering because their fixed prefix already guarantees low false positive rate.

---

Suppression

Inline — suppress a single line:

api_key = "AKIAIOSFODNN7EXAMPLE"  # nosecrets

File/directory — .secretsignore:

# gitignore-style syntax
tests/fixtures/
*.example
legacy_config.py

Copy .secretsignore.example from the repo as a starting point.

---

Exit codes

Code	Meaning
0	No findings at or above --fail-on threshold
1	One or more findings at or above threshold

Default --fail-on is high. Change with --fail-on critical to only block on critical findings.

---

CI integration

# GitHub Actions example
- name: Scan for secrets
  run: |
    pip install nosecrets
    secrets scan --json --severity medium

---

Philosophy

"A secrets scanner that leaks its own secrets is not a secrets scanner."

• Zero runtime dependencies — stdlib only, fully auditable
• Staged scanning reads from the git index (git show :<path>), not the working tree — you scan exactly what will be committed
• Secrets are never printed in full — output always redacts the middle portion
• Pattern quality over quantity — 20 well-tested patterns beat 500 noisy ones

---

License

MIT