Sleela Script Security
Sleela Script follows a capability-based host model.
Rules
- Core language evaluation is side-effect limited.
- Filesystem access requires a host capability.
- Process execution requires a host capability.
- Network access requires an explicitly registered capability.
- Native pointers are never script values.
- Host handles are opaque.
- Script modules resolve through approved search roots.
- Resource limits are host-controlled.
- Configuration is resolved through the canonical SLeeLa configuration root.
- Restricted CI mode should disable process, network, and arbitrary-write capabilities unless explicitly enabled.
A script being syntactically valid does not grant it native authority.
Resource limits
Hosts should be able to bound:
- execution time;
- instruction/evaluation steps;
- memory;
- recursion depth;
- output bytes;
- open handles;
- module count.
The exact mechanism is host-specific but the policy boundary is part of the language contract.