> For the complete documentation index, see [llms.txt](https://vayl.gitbook.io/vayl-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://vayl.gitbook.io/vayl-docs/documentation/why-vayl/auditable-gated-memory-for-high-stakes-ai-agents.md).

# Auditable, gated memory for high-stakes AI agents

**When an agent operates where a wrong answer causes harm — clinical, financial, legal — memory needs three things ordinary agent memory doesn't have: it must be correct under change, gated by a human on risky writes, and provable after the fact.** Vayl ships all three today.

## Correct under change

The reconciling core — one active value per fact, removal first-class, history preserved — so a superseded dose, plan, or clause never resurfaces as current.

## Gated by a human where it matters

* **Confirm-gated slots.** Declare fields (e.g. `active_medication`) so a change or removal is recorded as a *proposal*, not applied — a human `confirm_change` / `reject_change` decides, and *who decided* is recorded.
* **Critical-fact channel.** Tag categories (allergies, key covenants) `critical` so they **bypass ranking** and always surface — an omission here is a safety failure, not a quality one.
* **Act-time gates.** `check_before_act` / `safe_recall` withhold or block when a fact is disputed, stale, or low-confidence, instead of proceeding on shaky data.

## Provable after the fact

* **Signed decision snapshots.** `record_decision` binds an action to the exact facts the agent saw; `explain_decision` reconstructs it later, verified unaltered.
* **Tamper-evident audit chain.** Hash-chained, Ed25519-signed; `verify_audit` catches any edit, reorder, or truncation; anyone can verify offline with the public key.
* **GDPR built in.** `delete` / `delete_all` hard-erase with signed receipts; `export_memory` for DSARs; retention purges keep the chain verifiable.

## Honest boundaries

Vayl is secure-by-default for on-prem / self-hosted use, but it is **not independently audited or penetration-tested**, and makes no "compliant" guarantee — it gives you the *controls*; your DPO and security review sign off. It runs in your environment; nothing phones home.

## Next steps

<table data-view="cards"><thead><tr><th></th><th></th><th data-hidden data-card-target data-type="content-ref"></th></tr></thead><tbody><tr><td><h4><i class="fa-hospital" style="color:$primary;">:hospital:</i> Tutorial: a hospital assistant</h4></td><td>A gated, auditable clinical assistant, end to end.</td><td><a href="/pages/mgN0TUOebvjsXMBfpuQ6">/pages/mgN0TUOebvjsXMBfpuQ6</a></td></tr><tr><td><h4><i class="fa-gear" style="color:$primary;">:gear:</i> Safety gates &#x26; human approval</h4></td><td>Confirm gates, critical facts, and act-time checks in depth.</td><td><a href="/pages/d9d593aeabc52a165190bf8c93720491a4eb9682">/pages/d9d593aeabc52a165190bf8c93720491a4eb9682</a></td></tr><tr><td><h4><i class="fa-file-signature" style="color:$primary;">:file-signature:</i> Accountability</h4></td><td>Signed decisions, attestations, and the audit chain.</td><td><a href="/pages/R2NPTpA8ljGHbWMvp6xn">/pages/R2NPTpA8ljGHbWMvp6xn</a></td></tr></tbody></table>


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://vayl.gitbook.io/vayl-docs/documentation/why-vayl/auditable-gated-memory-for-high-stakes-ai-agents.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.