Documentation Index
Fetch the complete documentation index at: https://docs.getbifrost.ai/llms.txt
Use this file to discover all available pages before exploring further.
Overview
Secrets Detection is a Bifrost Enterprise guardrail provider that scans LLM request and response text for leaked credentials, API keys, access tokens, private keys, and other secret-shaped values. It runs entirely inside Bifrost. You do not need to configure an external moderation service, credentials, or provider account.Secrets Detection is for credential leakage. For personal data such as email addresses, phone numbers, SSNs, credit-card-like numbers, and IP addresses, use the Custom Regex PII Detection template.
How It Works
Bifrost uses the embedded default rules from Gitleaks v8.30.1 for thesecrets guardrail provider.
At runtime:
- You create a guardrail provider with
provider_name: "secrets". - You attach that provider to one or more guardrail rules.
- The rule decides when to run the provider and whether to scan
input,output, orboth. - Secrets Detection scans text blocks in the selected request/response phase.
- On the first detected secret, Bifrost returns
GUARDRAIL_INTERVENEDwith a reason such assecret detected: github-pat.
Supported Secret Types
The built-in rule set contains 222 Gitleaks default rules in this build. The coverage is broad, but the useful way to think about it is by credential family:| Family | Examples |
|---|---|
| Cloud provider credentials | AWS access keys and Bedrock API keys, Azure AD client secrets, GCP API keys, Alibaba Cloud keys, Yandex Cloud tokens, DigitalOcean tokens, Cloudflare keys |
| Source control and DevOps tokens | GitHub PATs/OAuth/App tokens, GitLab tokens, Bitbucket credentials, Atlassian tokens, Codecov, Drone CI, Travis CI, Harness, Sourcegraph |
| Package and artifact registry tokens | npm, PyPI, RubyGems, NuGet, Clojars, JFrog, Artifactory |
| AI and LLM provider keys | OpenAI, Anthropic, Cohere, Hugging Face, Perplexity, PrivateAI |
| Data, analytics, and monitoring keys | Databricks, Datadog, Dynatrace, Grafana, New Relic, Sentry, Sumo Logic, ClickHouse, Confluent, Algolia, Looker |
| Collaboration and messaging credentials | Slack tokens and webhooks, Discord tokens, Microsoft Teams webhooks, Telegram bot tokens, Mattermost, Gitter |
| Payments, finance, and commerce keys | Stripe, Square, Plaid, Coinbase, Kraken, KuCoin, Bittrex, Flutterwave, GoCardless, Shopify, Etsy |
| CRM, marketing, and support keys | HubSpot, Intercom, Zendesk, Mailchimp, Mailgun, SendGrid, Sendinblue, Contentful, Typeform |
| Infrastructure and secret material | HashiCorp Vault tokens, Terraform Cloud tokens, Kubernetes Secret YAML, OpenShift tokens, private keys, PKCS#12 files, Age secret keys, 1Password service account tokens, Doppler tokens |
| Generic credential patterns | Generic API keys, JWTs, base64 JWTs, curl authorization headers, curl basic auth user strings |
Current rule IDs
Current rule IDs
Configuration
- Web UI
- API
- config.json
- Helm
- Go to Guardrails > Providers.
- Select Secrets Detection.
- Click Add Configuration.
- Set a descriptive Name, for example
block-leaked-credentials. - Optionally add a False-positive allowlist.
- Enable the configuration and save it.
- Attach the configuration to a guardrail rule under Guardrails > Configuration.
False-Positive Allowlist
Theignored_secret_keywords setting is a list of substrings that should suppress known false positives.
Bifrost normalizes these values by trimming whitespace, lowercasing them, deduplicating them, and sorting them before loading the Gitleaks detector. A finding is skipped when the detected secret value contains one of those substrings.
Use this for stable test fixtures or placeholder values such as example, dummy, or a known internal sample-token prefix. Keep it narrow. A broad allowlist entry can hide real leaked credentials.
Operational Notes
- Secrets Detection has no external network dependency.
- It uses the default Gitleaks rule set embedded in the Enterprise build.
- It intervenes on the first detected secret in the evaluated text.
- Use guardrail rule
apply_toto control whether prompts, responses, or both are scanned. - Use rule-level
sampling_rateif you want to evaluate only a percentage of traffic. - Use Custom Regex for organization-specific patterns or PII templates that are not credential-focused.