Static copy

Copied from docs/en/governance/security-and-threat-model.md in https://github.com/poppy-team/aipo-lang (MIT). Pinned to revision ce47c4e070404fc41722ccd8e2b7b808b99062a8, blob 244a70c926a8bfe777c020bb14ad74a91fae4da6. The source repository remains canonical; this copy is refreshed through a sync pull request, not live.

Security & Sandboxing

Aipo security architecture is anchored in the principle of least privilege and proactive runtime vulnerability mitigation.

---

Security Principles

1. Zero Hardcoded Secrets: Source code, test fixtures, and package manifests never store credentials, API keys, or access tokens.
2. Deny by Default: No host system capability (filesystem, environment variables, system clock, network sockets) is accessible to scripts without explicit permission granted by the host application.
3. Memory Isolation Against Use-After-Free: Host resources exposed to Aipo scripts utilize generational handles with atomic slot retirement.
4. Denial of Service (DoS) Hardening:
   • Maximum recursion depth ceiling in the parser (256 levels) preventing call stack exhaustion.
   • Optional instruction step budgeting (gas budget) preventing infinite execution loops in untrusted scripts.
   • Varint LEB128 maximum byte length limits mitigating integer overflow attacks.
5. Supply Chain Hermeticity:
   • Remote package dependencies require explicit 40-character commit SHAs in aipo.toml.
   • Automatic cryptographic verification against SHA-256 tree digests stored in local cache.
   • Production builds execute 100% offline, eliminating dynamic runtime dependency poisoning.