Roadmap

We're building defence layers for AI agents. Here's what's live, what's coming next, and what's on the horizon.

Live in Production

Coming Soon

Planned

Content Scanning

LIVE

HTML hidden element detection
Base64/hex encoding detection
Unicode normalization
Shard attack detection
Multi-format parsing (Markdown, JSON)
Instruction pattern matching

Runtime Detection

LIVE

Behavior monitoring hooks
Output anomaly detection
Context hijacking detection
Privilege escalation alerts

Secret & Credential Scanning

LIVE

20 built-in credential types
Shannon entropy analysis for unknown secrets
Configurable actions (log, warn, redact, block)
Custom patterns and allowlist

Agent Integration

LIVE

LangChain adapter (SentinelAgentMiddleware + CallbackHandler)
CrewAI adapter (SentinelCrewAIMiddleware)
Haystack component (SentinelComponent)
AutoGen / AG2 adapter (SentinelAutoGenMiddleware)
Middleware/decorator patterns

Platform Adapters

PLANNED

OpenAI API wrapper
Anthropic Claude wrapper
HuggingFace inference adapter
Custom model endpoints

Standard Features

COMING

Real-time dashboard
Threat analytics
Custom rule builder
Team collaboration tools

Framework adapters shipped in v0.9.0

LangChain, CrewAI, Haystack, and AutoGen/AG2 adapters are live and installable now. Install all adapters with one command:

pip install "sentinel-security[frameworks]"

Or install individually: sentinel-security[langchain], sentinel-security[crewai], sentinel-security[haystack], sentinel-security[ag2]

Want to influence the roadmap?

We're building for teams running AI agents in production. If you have specific integration needs or feature requests, we want to hear from you.