Rules Reference
sekretbarilo ships 110 built-in detection rules that are enabled by default organized into three precision tiers. There are 113 total rules, including 3 optional public-key rules that must be opted in.
Three-Tier Detection System
Tier 1: Prefix-Based (75 rules)
Match distinctive, service-specific prefixes (e.g., AKIA, ghp_, sk-ant-). No keyword context needed. Very low false positive rate. Stopword filtering only checks for placeholder patterns like XXXX....
Tier 2: Context-Aware (32 rules)
Require keyword context (e.g., password=, postgres://) and/or entropy thresholds. Full stopword filtering. Password rules use strength heuristics instead of entropy.
Tier 3: Catch-All (3 rules)
Generic patterns with the highest entropy threshold (4.0). Keyword-oriented patterns cover api_key, auth_token, and similar names; generic-high-entropy-value has no keywords and uses bounded value-shape matching.
Tier 1: Prefix-Based Rules
Cloud Providers
| Rule ID |
Description |
Prefix |
aws-access-key-id |
AWS access key ID |
AKIA + 16 chars |
gcp-api-key |
GCP API key |
AIza + 35 chars |
gcp-oauth-client-secret |
GCP OAuth client secret |
GOCSPX- + 28 chars |
alibaba-access-key-id |
Alibaba Cloud access key ID |
LTAI + 12-20 chars |
digitalocean-personal-access-token |
DigitalOcean PAT |
dop_v1_ + 64 hex |
digitalocean-oauth-token |
DigitalOcean OAuth token |
doo_v1_ + 64 hex |
digitalocean-refresh-token |
DigitalOcean refresh token |
dor_v1_ + 64 hex |
Source Control
| Rule ID |
Description |
Prefix |
github-personal-access-token |
GitHub PAT |
ghp_ + 36+ chars |
github-oauth-token |
GitHub OAuth token |
gho_ + 36+ chars |
github-app-token |
GitHub app token |
ghs_ + 36+ chars |
github-refresh-token |
GitHub refresh token |
ghr_ + 36+ chars |
github-fine-grained-pat |
GitHub fine-grained PAT |
github_pat_ + 82+ chars |
gitlab-personal-access-token |
GitLab PAT |
glpat- + 20+ chars |
gitlab-pipeline-trigger-token |
GitLab pipeline trigger |
glptt- + 20+ chars |
gitlab-runner-registration-token |
GitLab runner registration |
glrt- + 20+ chars |
gitlab-ci-job-token |
GitLab CI job token |
glcbt- + 20+ chars |
Communication
| Rule ID |
Description |
Prefix |
slack-bot-token |
Slack bot token |
xoxb- + 24+ chars |
slack-user-token |
Slack user token |
xoxp- + 24+ chars |
slack-app-token |
Slack app token |
xapp- + 24+ chars |
discord-webhook-url |
Discord webhook URL |
discord.com/api/webhooks/... |
Payment
| Rule ID |
Description |
Prefix |
stripe-secret-key-live |
Stripe live secret key |
sk_live_ + 24+ chars |
stripe-secret-key-test |
Stripe test secret key |
sk_test_ + 24+ chars |
stripe-publishable-key-live |
Stripe live publishable key |
pk_live_ + 24+ chars |
stripe-restricted-key-live |
Stripe live restricted key |
rk_live_ + 24+ chars |
stripe-restricted-key-test |
Stripe test restricted key |
rk_test_ + 24+ chars |
square-access-token |
Square access token |
sq0atp- + 22+ chars |
square-oauth-secret |
Square OAuth secret |
sq0csp- + 40+ chars |
AI / ML
| Rule ID |
Description |
Prefix |
anthropic-api-key |
Anthropic API key |
sk-ant- + 20+ chars |
openai-api-key |
OpenAI API key (project) |
sk-proj- + 20+ chars |
openai-api-key-legacy |
OpenAI API key (legacy) |
sk-...T3BlbkFJ... |
huggingface-access-token |
HuggingFace token |
hf_ + 34+ chars |
replicate-api-token |
Replicate API token |
r8_ + 38+ chars |
Email / Messaging
| Rule ID |
Description |
Prefix |
sendgrid-api-key |
SendGrid API key |
SG. + base64 segments |
mailgun-private-api-token |
Mailgun private API token |
key- + 32 hex |
mailchimp-api-key |
Mailchimp API key |
32 hex + -us + digits |
sendinblue-api-key |
Brevo (Sendinblue) API key |
xkeysib- + 64 hex |
CI / CD
| Rule ID |
Description |
Prefix |
buildkite-api-token |
Buildkite API token |
bkua_ + 40 hex |
terraform-cloud-token |
Terraform Cloud token |
14 chars + .atlasv1. + 60+ chars |
Monitoring / Observability
| Rule ID |
Description |
Prefix |
new-relic-api-key |
New Relic API key |
NRAK- + 27 chars |
grafana-cloud-api-token |
Grafana Cloud API token |
glc_ + 32+ chars |
grafana-service-account-token |
Grafana service account token |
glsa_ + 32+ chars |
sentry-auth-token |
Sentry auth token |
sntrys_ + 36+ chars |
sentry-dsn |
Sentry DSN URL |
https://...ingest.sentry.io/... |
dynatrace-api-token |
Dynatrace API token |
dt0c01. + 24+64 chars |
Database
| Rule ID |
Description |
Prefix |
database-connection-string-postgres |
PostgreSQL connection string |
postgres:// or postgresql:// |
database-connection-string-mysql |
MySQL connection string |
mysql:// |
database-connection-string-mongodb |
MongoDB connection string |
mongodb:// or mongodb+srv:// |
redis-connection-string |
Redis connection string |
redis:// |
planetscale-password |
PlanetScale password |
pscale_pw_ + 30+ chars |
planetscale-api-token |
PlanetScale API token |
pscale_tkn_ + 30+ chars |
Package Registries
| Rule ID |
Description |
Prefix |
npm-access-token |
npm access token |
npm_ + 36+ chars |
pypi-api-token |
PyPI API token |
pypi- + 16+ chars |
docker-hub-pat |
Docker Hub PAT |
dckr_pat_ + 24+ chars |
rubygems-api-key |
RubyGems API key |
rubygems_ + 48 hex |
nuget-api-key |
NuGet API key |
oy2 + 43 chars |
Crypto / Secrets Management
| Rule ID |
Description |
Prefix |
pem-private-key |
PEM private key block |
-----BEGIN...PRIVATE KEY----- |
pgp-private-key-block |
PGP private key block |
-----BEGIN PGP PRIVATE KEY BLOCK----- |
jwt-token |
JWT token |
eyJ...eyJ... (3 base64 segments) |
age-secret-key |
age encryption secret key |
AGE-SECRET-KEY-1 + 58 chars |
hashicorp-vault-service-token |
Vault service token |
hvs. + 24+ chars |
hashicorp-vault-batch-token |
Vault batch token |
hvb. + 24+ chars |
Cloud Infrastructure
| Rule ID |
Description |
Prefix |
fly-io-api-token |
Fly.io API token |
fo1_ + 40+ chars |
pulumi-access-token |
Pulumi access token |
pul- + 40 hex |
SaaS
| Rule ID |
Description |
Prefix |
linear-api-key |
Linear API key |
lin_api_ + 40+ chars |
shopify-access-token-admin |
Shopify admin token |
shpat_ + 32 hex |
shopify-access-token-custom-app |
Shopify custom app token |
shpca_ + 32 hex |
shopify-access-token-private-app |
Shopify private app token |
shppa_ + 32 hex |
sourcegraph-access-token |
Sourcegraph access token |
sgp_ + 40+ hex |
figma-personal-access-token |
Figma PAT |
figd_ + 40+ chars |
mapbox-api-token |
Mapbox API token |
pk. + base64 segments |
dropbox-api-token |
Dropbox API token |
sl. + 100+ chars |
launchdarkly-sdk-key |
LaunchDarkly SDK key |
sdk- + UUID |
notion-api-token |
Notion API token |
ntn_ + 40+ chars |
databricks-api-token |
Databricks API token |
dapi + 32 hex |
facebook-access-token |
Facebook access token |
EAA + 20+ chars |
Tier 2: Context-Aware Rules
These rules require keyword context and apply additional validation (entropy thresholds and/or password strength heuristics).
Cloud & Infrastructure
| Rule ID |
Keywords |
Entropy |
aws-secret-access-key |
aws_secret, secret_access_key |
3.5 |
azure-storage-account-key |
accountkey |
— |
azure-ad-client-secret |
azure, client_secret |
3.5 |
azure-devops-pat |
azure, devops |
3.5 |
alibaba-secret-key |
alibaba, aliyun |
3.5 |
cloudflare-api-key |
cloudflare, cf_api |
3.0 |
heroku-api-key |
heroku |
3.0 |
Communication
| Rule ID |
Keywords |
Entropy |
discord-bot-token |
discord, bot |
3.0 |
telegram-bot-token |
telegram, bot |
3.0 |
twilio-api-key |
twilio |
— |
webhook-url-with-token |
hooks.slack.com |
— |
Databases
| Rule ID |
Keywords |
Entropy |
mssql-connection-string |
server, data source, password |
3.0 |
airtable-api-key |
airtable |
— |
Monitoring / Observability
| Rule ID |
Keywords |
Entropy |
datadog-api-key |
datadog, dd_api |
3.0 |
elastic-api-key |
elastic |
3.5 |
splunk-hec-token |
splunk |
3.0 |
pagerduty-api-key |
pagerduty |
3.0 |
CI / CD & DevOps
| Rule ID |
Keywords |
Entropy |
circleci-api-token |
circleci |
3.0 |
vercel-api-token |
vercel |
3.5 |
netlify-access-token |
netlify |
3.5 |
gitlab-deploy-token |
gitlab, deploy_token |
3.0 |
Identity & Auth
| Rule ID |
Keywords |
Entropy |
okta-api-token |
okta |
3.0 |
atlassian-api-token |
atlassian, jira |
3.5 |
twitter-bearer-token |
twitter, bearer |
3.5 |
http-bearer-token |
bearer, authorization |
3.5 |
http-basic-auth |
basic, authorization |
3.0 |
Email
| Rule ID |
Keywords |
Entropy |
postmark-server-token |
postmark |
3.0 |
AI / ML
| Rule ID |
Keywords |
Entropy |
cohere-api-key |
cohere |
3.5 |
Search
| Rule ID |
Keywords |
Entropy |
algolia-api-key |
algolia |
3.0 |
Generic Patterns
| Rule ID |
Keywords |
Entropy |
generic-password-assignment |
password, passwd, pwd |
strength heuristic |
generic-secret-assignment |
secret, secret_key, api_secret |
3.5 |
password-in-url |
:// |
strength heuristic |
Password rules (generic-password-assignment, password-in-url) only flag strong passwords: 8+ chars, mixed case, digits.
generic-password-assignment detects password, passwd, and pwd assignments with = or :, including names such as DB_PASSWORD and quoted mapping keys. Values can be double-quoted, single-quoted, backtick-quoted, or unquoted. Quoted captures preserve internal spaces, opposite quote characters, and escaped quotes. Unquoted captures stop at whitespace or syntax delimiters such as commas, semicolons, brackets, and parentheses; escaped whitespace and delimiters are included. Quote passwords containing literal delimiters. Masking replaces the complete captured value while retaining its surrounding quotes and adjacent fields.
All forms retain the same rule ID and its value allowlists and password-strength filters. Shell variables, supported template expressions, and weak or placeholder passwords remain excluded. This is pattern detection rather than a parser for every configuration language.
Raw single-quoted and backtick strings can be ambiguous with languages that interpret backslash escapes. When a backslash precedes a possible closing quote and another matching quote follows, detection conservatively includes the longer interpretation. This can mask adjacent text in raw-string formats. Multiline quoting and mixed shell quoting are not parsed as password assignments.
Tier 3: Catch-All Rules
| Rule ID |
Keywords |
Entropy |
generic-api-key |
api_key, apikey, api-key, api_token, api-token |
4.0 |
generic-token-assignment |
auth_token, access_token, secret_token |
4.0 |
generic-high-entropy-value |
none |
4.0 |
generic-high-entropy-value detects a complete single-line ASCII token value when the captured value is at least 20 bytes and its raw, case-sensitive byte Shannon entropy is at least 4.0 bits per byte. It scores only the captured value, excluding the variable or key name, delimiter, and surrounding quotes; it does not normalize case or adjust for charset size. Whitespace-containing values and complete variable references are not candidates.
Supported assignment forms are NAME=VALUE, export NAME=VALUE, NAME: VALUE, name = “VALUE”, name = ‘VALUE’, and JSON “name”: “VALUE”, including multiple fields on one line. A standalone otherwise-bare alphanumeric, base64, or base64url token line also qualifies, including normal trailing base64 padding and optional surrounding single or double quotes and horizontal whitespace.
The variable or key name is irrelevant: there are no exemptions for PATH, MANPATH, LS_COLORS, TERM_SESSION_ID, DB_TOKEN, or similar names. This intentional broadness can flag harmless high-entropy data such as base64 blobs and lockfile-style checksums. To suppress a specific false positive, add a matching value to [allowlist].stopwords, or preferably add a [[allowlist.rules]] entry for generic-high-entropy-value with regexes matched against the captured value. For known-safe assignment names, its per-rule keys allowlist provides case-sensitive whole-key matching without weakening other rules.
This rule bypasses the built-in default stopword list, surrounding-line hash/checksum suppression, and documentation-file entropy bonus used by other rules. Explicit user-configured [allowlist].stopwords still apply as case-insensitive plain substrings of the captured value, not word-boundary matches. A per-rule [[allowlist.rules]] value regex for generic-high-entropy-value is narrower and more precise; anchor it to an exact known-safe value instead of using a broad pattern. Path allowlists, public-key suppression, and a higher global [settings] entropy-threshold override still apply normally; the override is a floor that can raise, never lower, this rule’s 4.0 threshold.
Gated Rules: Public Key Detection
these 3 rules are disabled by default to reduce noise. enable them with detect_public_keys = true in config or --detect-public-keys on the CLI.
| Rule ID |
Description |
Matches |
pem-public-key |
PEM public key header |
-----BEGIN PUBLIC KEY-----, -----BEGIN RSA PUBLIC KEY-----, etc. |
pgp-public-key-block |
PGP public key block header |
-----BEGIN PGP PUBLIC KEY BLOCK----- |
openssh-public-key |
OpenSSH public key |
ssh-rsa AAAA..., ssh-ed25519 AAAA..., ecdsa-sha2-nistp256 AAAA..., etc. |
when disabled (default), sekretbarilo also suppresses false positives from base64 content inside multi-line PEM/PGP public key blocks (e.g., base64 lines that might trigger token rules like facebook-access-token).
enabling public key detection:
# .sekretbarilo.toml
[settings]
detect_public_keys = true
or via CLI:
sekretbarilo scan --detect-public-keys
sekretbarilo audit --detect-public-keys
False Positive Reduction
- Entropy thresholds — tier 2/3 rules filter low-randomness strings; doc files get +1.0 bonus (the bonus does not apply to
generic-high-entropy-value)
- Stopwords —
example, test, placeholder, changeme, fake, mock, dummy, etc. (generic-high-entropy-value uses only explicit user-configured stopwords)
- Hash detection — SHA-1, SHA-256, MD5, git commit hashes (
generic-high-entropy-value bypasses this suppression)
- Variable references —
${VAR}, process.env.VAR, os.environ["VAR"], etc.
- Template handling — Jinja2/Helm/Mustache/Handlebars ``, GitHub Actions
$, ERB <%= %>, Terraform ${var.}, etc. (generic-high-entropy-value bypasses this suppression)
- Public key suppression — PEM, PGP, and OpenSSH public key blocks are suppressed by default (prevents base64 content from triggering token rules)
- Password strength — only flags strong passwords (8+ chars, mixed case, digits); also applied to connection string passwords
- Path allowlists — binary, generated, lock files, vendor dirs auto-skipped
Custom Rules
Add project-specific rules in .sekretbarilo.toml:
[[rules]]
id = "custom-internal-token"
description = "Internal service token"
regex = "(MYCO_[A-Z0-9]{32})"
secret_group = 1
keywords = ["myco_"]
entropy_threshold = 3.5
[rules.allowlist]
regexes = ["test_token_.*"]
paths = ["test/.*"]
Required fields
id — unique identifier (lowercase with hyphens)description — human-readable descriptionregex — pattern with capture group for the secretsecret_group — which capture group contains the secret (usually 1)keywords — case-insensitive keywords for aho-corasick pre-filtering
Optional fields
secret_groups — alternative capture group indices, checked in order if secret_group did not participate. Defaults to []. If no configured group participates, the full match is used. Overrides replace the complete rule, including this list.entropy_threshold — minimum Shannon entropy (typical: 3.0–4.0)allowlist.regexes — value patterns to skipallowlist.paths — file path patterns to skip
Override or disable built-in rules
Override by defining a rule with the same id. Disable with an unmatchable pattern:
[[rules]]
id = "generic-api-key"
description = "Disabled"
regex = "(?-u:^$a)"
secret_group = 1
keywords = ["__never_match__"]
Configuration hierarchy
Rules merge in this order (later overrides earlier):
- Built-in defaults (113 definitions, 110 enabled by default)
- System config (
/etc/sekretbarilo.toml)
- User config (
~/.config/sekretbarilo.toml)
- Project config (
.sekretbarilo.toml)
- CLI overrides (
--config, --no-defaults, etc.)
Same id replaces the earlier definition; unique ids are appended.
Next Steps