Pay for what you intercept.
No per-seat fees. No data egress costs. Start free, scale when your AI agents do.
Hacker
For indie devs exploring AI security.
5,000 intercepts / month
- Prompt, command & diff scanning
- Built-in secret detection (AWS, GitHub, PII...)
- MAS-TRM risk scoring
- 3 custom regex rules
- 10 MCP Copilot requests / month
- 7-day audit log retention
- Community support
Pro Developer
For teams shipping AI-powered products.
150,000 intercepts / month
- Everything in Hacker
- Unlimited custom interceptor rules
- Regex tester with inline feedback
- 50 MCP Copilot requests / month
- 30-day audit log retention
- Block records & dedup hashing
- Priority email support
Business
For companies running AI at scale.
1,000,000 intercepts / month
- Everything in Pro
- 1M intercepted requests / month
- 200 MCP Copilot requests / month
- 90-day audit log retention
- Multi-seat team access
- GitHub App — commit interception
- Dedicated Slack support channel
Enterprise
For regulated industries and large teams.
Unlimited
- Everything in Business
- Dedicated single-tenant deployment
- Self-hosted VPC option (Rust binary)
- Custom SLA with 99.99% uptime
- 500 MCP Copilot requests / month
- 1-year log retention
- SSO / SAML integration
- White-glove onboarding
Need more intercepts mid-cycle?
Buy top-up API credits on demand. $10 = 10,000 requests ($0.001 per intercept). Minimum $10.
Need more AI Copilot interactions?
Buy top-up MCP Copilot credits. $5 = 100 requests ($0.05 per AI request). Minimum $5.
What's in every plan
Prompt scanning
Detect hardcoded secrets, PII, DB connection strings, and private keys in LLM prompts.
Command scanning
Block destructive file deletion, privilege escalation, raw disk writes, and unsafe pipelines.
Diff scanning
Scan AI-generated code diffs for credentials, tokens, and Slack webhooks before git commit.
MAS-TRM risk scoring
Every finding is scored on a Likelihood × Impact matrix. Results rated Low → Critical.
Cryptographic audit trail
Every blocked action generates an immutable, searchable audit record.
Zero-knowledge storage
Payload content is never written to disk. Only SHA-256 hashes for deduplication.
Frequently asked
What does BLAZLE actually intercept?
Three surfaces: (1) LLM prompts — checked for secrets, PII, and injections. (2) Shell commands — checked for destructive ops, privilege escalation, and unsafe pipelines. (3) Git diffs — checked for hardcoded credentials and tokens.
What counts as one intercept?
Each call to /intercept/prompt, /intercept/command, or /intercept/diff counts as one request against your monthly quota.
How is risk scored?
Every finding is mapped to a Likelihood × Impact matrix derived from MAS-TRM (Monetary Authority of Singapore Technology Risk Management guidelines). Scores above 75 are blocked by default.
Is my payload stored?
Never. Payload content is processed entirely in-memory and never written to disk. Only the SHA-256 hash of blocked payloads is retained for deduplication. Zero-knowledge storage by design.
What happens when I exceed my quota?
By default, requests pass through unscanned when your quota is exhausted. You can enable overage billing or buy PAYG credits at any time from your dashboard.
Can I self-host BLAZLE?
Enterprise customers can request the compiled Rust binary for self-hosted deployment within their own VPC or on-premise infra. Reach out to discuss.
Start protecting your AI agents today.
Free forever on the Hacker tier. No credit card required.