Security and privacy

Principles

Least privilege, read-only by default, untrusted-input handling, no secrets in artifacts, honest limits.

Threat table

ThreatControlTest
Malicious / oversized ONNX32 MiB cap, field/node/opset caps, bounded varints, no executiontests/onnx.test.mjs (truncated, malformed wire, giant payload, UTF-8)
Parser crash on corrupt inputAll failures are plain Error; mutation fuzz500-case test + 2000-case proof run, 0 unexpected crashes
Symlink read/write redirectionlstat refusal for JSON, ONNX, reports and installer destinationssymlink tests (skip only where the OS forbids symlinks)
XSS via model / manifest text in reportsEscape every interpolated valueHTML escapes external content, hostile audit test
Report tamperingSHA-256 recomputation (verify, render refuse invalid)tamper tests
Token theft / brute force on serve≥16-char token, constant-time compare, 127.0.0.1 onlyserver tests
Path traversal on serveFilename allowlist regex, .. rejectiontraversal test
Overwriting user files on installNo-overwrite policyoverwrite test
Supply-chainZero dependencies; CI secret scan; npm provenance on releaseCI
Prompt injection from model cardsSkills treat all model metadata as untrusted; no execution of model-provided commandsskill text, rules/input-trust.md

Privacy

No telemetry or network calls. Hostname is reduced to a 10-hex-char SHA-256 prefix. No serials, usernames, or environment variables are collected. Files written with 0600 on POSIX.

Known limits (not hidden)

Reporting

GitHub private vulnerability reporting on the repository.