UVLM Trauma Informed AI Protocol (TIAIP) Governance Module for CoherenceLattice · v0.1.0-alpha.3

Public open-source alpha — preconformance

UVLM Trauma Informed AI Protocol (TIAIP) Governance Module for CoherenceLattice

A stateless, inspectable Trauma Informed AI Protocol sidecar for trauma-sensitive candidate review, bounded repair, and human authority.

Use selected trauma-informed safeguards without turning a model into a clinician, survivor classifier, crisis service, investigator, or final decision-maker.

Local and stateless
No model required
No network
No persistent memory
Human repair authority
Authority effect: NONE

Download alpha.3 Read how the thin module works

Trauma-sensitive safeguards without deciding who is traumatized

The module's central rule is simple: behave competently when trauma may be relevant; do not infer, diagnose, rank, or classify a person as traumatized.

Preserve

Keep voice, uncertainty, and source boundaries visible

First-person position, user choice, source-as-data separation, candidate quarantine, and exact lineage remain explicit.

Review

Run bounded native checks

The sidecar checks safe-scope partitioning, privacy and claim patterns, forced proximity, qualified-human routing, immediate-safety declarations, and authority laundering.

Repair

Require a named human decision

One exact REQUEST_REPAIR can authorize one bounded successor. The candidate cannot repair, approve, publish, or deploy itself.

Thin sidecar, not the standalone TIAIP application. This release does not include a browser UI, ingestion stack, connector, persistent store, export service, installer, deletion service, federation, or model runtime.

Why can a small module do real work? Read “Small by Design” for the complete architecture explanation.

What the reference workflow demonstrates

The package executes a complete local evidence route without calling a model or network service.

Validate one typed request. Source material is data and cannot grant authority.
Build a deterministic preflight. The module derives the active scope and required checks from trusted inputs.
Quarantine one candidate. Candidate identity is bound before substantive review.
Emit a native finding set. The reference candidate opens six expected findings.
Record human REQUEST_REPAIR. The instruction and permitted surfaces are exact.
Issue one consume-once repair packet. Reuse, unrelated rewrite, and authority escalation fail closed.
Validate one successor. The reference successor closes the six findings without silently overwriting the predecessor.
Seal telemetry and a replay receipt. The receipt records process identity; it does not certify truth or safety.

Controlling public download

The exact alpha.3 source, checksum, validation, release receipt, and separate-pass review are publicly hosted. Verify the source hash before use.

Exact source archive

UVLM_TIAIP_CoherenceLattice_Module_v0.1.0-alpha.3.zip

151,516 bytes · 94 members

1c33aefbf4faef3497b1efac4825dfe53cc8d5e8e18b35c6f7cfdab5bc6f4b58

Download exact source ZIP Download SHA-256

Verify first

Compare the downloaded bytes

Windows PowerShell:

Get-FileHash .\UVLM_TIAIP_CoherenceLattice_Module_v0.1.0-alpha.3.zip -Algorithm SHA256

Linux or macOS:

sha256sum UVLM_TIAIP_CoherenceLattice_Module_v0.1.0-alpha.3.zip

Supported claim and hard ceiling

This alpha keeps one bounded promise.

Supported public claim. This source alpha adds inspectable trauma-sensitive interaction safeguards aligned with selected trauma-informed principles within a bounded research claim.

It can help preserve

First-person speaker position and user choice.
Uncertainty and source-as-data separation.
Candidate quarantine, typed findings, and bounded repair lineage.
No automatic contact, no forced proximity, and human final authority.

It does not establish

Trauma status, diagnosis, credibility, dangerousness, incapacity, worth, or identity.
Clinical, legal, crisis, safeguarding, privacy, de-identification, or truth certification.
Production readiness, universal safety, model empathy, or prevention of retraumatization.
Exact Product Module Contract alpha.6 conformance.

No high-impact use. This source alpha is not authorized for clinical, legal, safeguarding, employment, housing, credit, custody, benefits, policing, immigration, emergency, or other consequential automated decisions.