Behavioral runtime security for web applications.
Tellsign is coming.
TLDR: Anomalous behavior logged or prevented.
Basically, a whitelist of acceptable functionality.
Anything else gets logged or guarded against.
Real world example:
A malicious user starts taking liberties with application inputs and tries to inject own parameters to force the application to follow new paths of execution that it was not designed for.
In the wild, this would isually mean identifying a bug and trying to exploit it.
This event would require investigation or even countermeasures.
Tellsign learns how your application actually executes during testing — every method call, argument type, and execution path — then watches production for deviations.
When runtime behavior diverges from the trained baseline, it flags it.
The deviation is either a gap in your test coverage or an attacker doing something your developers never anticipated. Both are valuable signals.
In production, every request is traced.
New execution paths trigger a log event.
Configurable via TOML: inclusion and exclusion rules, custom value types, pattern matching, extensions.
Logs can be piped to any LLM provider for analysis using the customer's own API tokens.
Optional Guard mode errors out on anomalous behavior, turning detection into prevention.
Cheers!