Provenance and governance
Evidence URLs, the transparency log, and the classification gate that keeps secrets in.
Provenance answers "can I trust this artifact?". Governance answers "is it safe to share?". SEIF treats both as first-class.
Evidence and verification
Signing produces an Evidence URL. The record it points to carries:
| Field | Meaning |
|---|---|
| Signature | Ed25519 signature over the content hash |
| Hash | SHA-256 of the exact bytes signed |
| Timestamp | OpenTimestamps anchor (independently verifiable) |
| Class | PUBLIC, INTERNAL, or CONFIDENTIAL |
Verification does not require SEIF's servers to be online — the cryptographic chain stands on its own.
seif verify <evidence-url>The transparency log
Signed artifacts are appended to a transparency log compatible with Certificate-Transparency-style auditing. The log is append-only and supports inclusion proofs, so the history cannot be quietly rewritten.
The classification gate
Before any content leaves a workspace, it passes the classification gate.
PUBLIC
Open evidence, verifiable by anyone with the URL. Default for open work.
INTERNAL
Workspace-scoped. Evidence is visible only to authenticated members.
CONFIDENTIAL
Hash-anchor only in the public log; the full record needs authenticated retrieval.
Escalation is one-way
Classification only escalates, never silently downgrades. Markers such as
password, token, api_key, secret, and CVE auto-escalate to
CONFIDENTIAL. Promotion to PUBLIC requires explicit human confirmation.
Just-in-time expertise
SEIF makes the AI a just-in-time expert in its own capabilities, so the human never has to be. Surfacing a compact capability map lifted a cold model's verb-reach from 0.50 to 1.00 — measured.
Quality and grounding
The form quality gate (A-F grade) was removed in c754. SEIF's verification value now comes from provenance and §7 enforcement — not form-based scoring.