CAIN Proof-Carrying Code, explained for beginners
PCC refuses agent-written code that could run shell commands or arbitrary code, before it runs.
What it is
Agents write and run code. PCC checks that code first, by reading its structure (its syntax tree), not by running it.
Code that parses, imports only allowed modules and has no dynamic execution is CERTIFIED_SAFE and lists the checks it passed. Code that calls os.system, eval and the like fails verification.
An everyday comparison
A building inspector who reads the plans and refuses any with a door to the gas main.
How it works, in 4 steps
- The agent submits code.
- Its structure is analysed without running it.
- All checks pass: certified. Any fails: refused.
- The result lists each check.
Try it now
Open the three live demo tools → Every button sends a real request to the live product and shows the real answer.
See it on the homepage: CAIN Proof-Carrying Code section →
Who it is for
Teams letting agents generate and run code.
FIRME AI CORPORATION · CAIN-42 is a Byzantine trust fabric. Numbers marked "our own check" were run by us, not by an independent party. Nothing on these pages is a certification.