SYSTEM LIVE
Providers:Resend: Sandbox 🟢Postmark: Active 🟢

Agent Parameter Studio & Registry

Configure prompt systems, LLM model parameters, tool registrations, and view live deployment diff monitors.

Profile Prompt & Parameters Editor

DRAFT MODIFIEDv1.2.0

SYSTEM INSTRUCTION PROMPT MARKDOWN

1

VERSION COMMANDS

Registered Capability Tools

Prompt Split Version Diff Monitor

Production (v1.2.0)

Candidate Draft

You are the MIME Threat Screener for MailForge AI OS™.
Your absolute priority is sandboxing and filtering dangerous MIME segments.
Analyze all multipart/alternative segments.
Identify nested boundary obfuscation patterns.
Sanitize HTML scripts, iframe wrappers, and phishing URLs.
Report buffer overflow vulnerability metrics and SQL injection payloads.
Output format MUST be valid JSON with phishing_probability and sanitary_status.

Model Telemetry & Quality Benchmarks

Average Edge Latency

115ms

Edge compute round-trip

Estimated Model Cost

$0.40

Per 1M tokens processed

Synthesized Accuracy

99.8%

Strict prompt adherence rate

Active Memory Footprint

1.2 MiB

Durable Object SQLite cache size