Security as a verifiable property.
Capabilities declared. Kernel-enforced. Signed.
Locara apps run with elevated capabilities on user
machines. Security is the project's first principle. Every
capability is declared in the manifest, every model is
pinned by SHA-256, every tool runs in a Wasmtime sandbox,
and every .locapp is signed by Locara CI with
a provenance attestation.
Every .locapp verifies itself before it runs.
The runtime refuses anything it can't verify — a missing
signature, a model whose hash doesn't match, a capability
the manifest didn't declare. The output below is what the
CLI prints when you run locara verify on the
Transcribe reference app.
- ✓ Manifest schema v1.0, well-formed
- ✓ Capabilities 4 grants, kernel-enforced
- ✓ Model integrity whisper-base.en@sha256:9b3a…
- ✓ Tool sandbox WASM, sealed
- ✓ Network access denied (not declared)
- ✓ Binary signature locara-ci · 2026-05-03
Reporting a vulnerability.
Email security@locara.app. Please do not open a public issue.
PGP fingerprint will be published here once the trust group keys are generated. We aim to acknowledge reports within 72 hours and ship a coordinated fix in line with each bug's severity.
Where the security model is documented.
Coordinated disclosure log.
None published yet. When advisories exist, they will be listed here with their CVE / GHSA identifiers, affected versions, mitigations, and timeline.