> 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/reference/mcp-tools.md).

# MCP tools

All 32 MCP tools Vayl exposes, grouped by job, with the capability each one needs.

Vayl's API is a set of 32 MCP tools. There is no REST layer: any MCP client connects over stdio (`vayl-mcp`) or authenticated HTTP (`vayl-server`) and calls them by name. Usually the LLM in your agent picks the tool; you can also call them from your own code, see [Calling Vayl from code](/vayl-docs/documentation/getting-started/calling-vayl-from-code.md).

## Reading these pages

Every tool section shows:

* the **signature** with its real defaults, copied from `src/vayl/api/mcp_server.py`;
* the **capability** a caller's role must grant (see [Capabilities](#capabilities));
* the tool's **MCP annotations** (read-only, destructive, open-world);
* a **real return string**, captured by running the tool offline with the model stubbed. Where the text comes from the answering model (`recall`, `recall_related`, `safe_recall` when it answers, the `based on:` line of `record_decision`), the page says so and the example is representative only.

Over MCP you pass arguments as a JSON object:

```json
{"name": "remember", "arguments": {"text": "We use Postgres", "user_id": "proj_7"}}
```

For the exact, always-current schema of every tool, call `tools/list` (see [The MCP interface](/vayl-docs/documentation/reference/mcp-tools/the-mcp-interface.md)).

## Memory-space arguments

A [memory space](/vayl-docs/documentation/concepts/memory-spaces.md) is the triple `(user_id, agent_id, run_id)`. Tools that read or write memory take those three arguments, defaulting to `user_id="default"`, `agent_id=""`, `run_id=""`. Not every tool does:

| Takes `user_id`, `agent_id`, `run_id`                                                                                                                                                                                                                                                                                                       | Takes only `user_id`                                                                    | Takes none                                                                                                                                            |
| ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------- |
| `remember`, `recall`, `recall_related`, `forget`, `get_memory`, `update_memory`, `history`, `list_memories`, `record_decision`, `check_before_act`, `safe_recall`, `pending_changes`, `confirm_change`, `reject_change`, `set_reconcile_policy`, `get_reconcile_policy`, `attest`, `delete`, `delete_all`, `export_memory`, `purge_expired` | `explain_decision` (default `"default"`), `audit_log` (default `""` = the whole tenant) | `verify_receipt`, `verify_audit`, `export_public_key`, `create_principal`, `list_principals`, `revoke_principal`, `license_status`, `stats`, `health` |

## Capabilities

On `vayl-server` every call is checked before the tool runs, and denials are written to the audit log. Over stdio (`vayl-mcp`) the caller is a trusted local admin, unless `VAYL_AUTH_REQUIRED` is on.

| Capability | Roles that have it           | Tools                                                                                                                                                               |
| ---------- | ---------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `read`     | admin, member, agent, viewer | `recall`, `recall_related`, `get_memory`, `history`, `list_memories`, `check_before_act`, `safe_recall`, `pending_changes`, `get_reconcile_policy`, `export_memory` |
| `write`    | admin, member, agent         | `remember`, `forget`, `update_memory`, `record_decision`, `attest`                                                                                                  |
| `delete`   | admin, member                | `delete`                                                                                                                                                            |
| `verify`   | all five roles               | `explain_decision`, `verify_receipt`, `verify_audit`, `export_public_key`, `audit_log`, `license_status`, `stats`, `health`                                         |
| `approve`  | admin, member                | `confirm_change`, `reject_change`                                                                                                                                   |
| `admin`    | admin                        | `delete_all`, `purge_expired`, `set_reconcile_policy`, `create_principal`, `list_principals`, `revoke_principal`                                                    |

Three tools add a check of their own: `audit_log` without a `user_id` needs `admin`; `verify_receipt` only verifies receipts owned by a space in the caller's scope; `remember` with a trusted `source` needs a key named after that source, or `approve`. On a multi-tenant deployment, some admin tools also depend on the caller's tenant: see [The deployment operator](/vayl-docs/documentation/concepts/authentication-and-access.md#the-deployment-operator). See [Authentication & access](/vayl-docs/documentation/concepts/authentication-and-access.md) for roles, scopes and tenants.

## Groups

| Group                                                                                  | Tools                                                                                                                                   |
| -------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------- |
| [Memory](/vayl-docs/documentation/reference/mcp-tools/memory.md)                       | `remember`, `recall`, `recall_related`, `forget`, `history`, `list_memories`, `get_memory`, `update_memory`                             |
| [Safety and gating](/vayl-docs/documentation/reference/mcp-tools/safety-and-gating.md) | `check_before_act`, `safe_recall`, `pending_changes`, `confirm_change`, `reject_change`, `set_reconcile_policy`, `get_reconcile_policy` |
| [Accountability](/vayl-docs/documentation/reference/mcp-tools/accountability.md)       | `record_decision`, `explain_decision`, `attest`, `verify_receipt`, `audit_log`, `verify_audit`, `export_public_key`                     |
| [Compliance (GDPR)](/vayl-docs/documentation/reference/mcp-tools/compliance-gdpr.md)   | `delete`, `delete_all`, `export_memory`, `purge_expired`                                                                                |
| [Administration](/vayl-docs/documentation/reference/mcp-tools/administration.md)       | `create_principal`, `list_principals`, `revoke_principal`, `license_status`, `stats`, `health`                                          |

## 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-plug" style="color:$primary;">:plug:</i> The MCP interface</h4></td><td>Transports, <code>tools/list</code>, the call envelope, annotations, and errors.</td><td><a href="/vayl-docs/documentation/reference/mcp-tools/the-mcp-interface.md">The MCP interface</a></td></tr><tr><td><h4><i class="fa-database" style="color:$primary;">:database:</i> Memory</h4></td><td>The core store-and-recall tools.</td><td><a href="/vayl-docs/documentation/reference/mcp-tools/memory.md">Memory</a></td></tr><tr><td><h4><i class="fa-lock" style="color:$primary;">:lock:</i> Authentication &#x26; access</h4></td><td>Roles, capabilities, scopes, and tenants.</td><td><a href="/vayl-docs/documentation/concepts/authentication-and-access.md">Authentication and access</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 by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://vayl.gitbook.io/vayl-docs/documentation/reference/mcp-tools.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

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.