agent-vault-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| AGENT_VAULT_AGENT_ID | No | Host-declared audit identifier for the agent. This is not strong authentication. | |
| AGENT_VAULT_SERVER_URL | No | Optional remote Agent Vault Service URL. By default this is unset, and the user chooses local or remote on first use. A saved/confirmed target takes precedence over this environment default. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| vault_statusC | Connection state and safe browser link; never secrets. |
| vault_select_targetB | Select local or probe a user-specified HTTPS vault server. No database is created and no old session is replaced until browser authentication succeeds. |
| vault_listB | List credential metadata only. |
| vault_findC | Find credential metadata by service. |
| vault_begin_enrollmentC | Request credential entry in the browser. Never pass secrets. |
| vault_resultC | Read enrollment state/reference only. |
| vault_declineC | Decline an enrollment. |
| list_operationsC | List controlled operations supported by the selected backend. |
| list_connectionsC | List connection metadata only. |
| prepare_requestC | Create a controlled operation ticket. |
| execute_operationC | Execute a registered operation inside the local core; remote legacy consume is not exposed. |
| vault_export_credentialA | Write a stored credential to a protected local file and return only the file path. The secret never appears in this result; an agent with file access can read the file afterwards. format=json writes {service, account, secret} for login-style use. |
| get_connection_statusC | Get connection metadata. |
| get_operation_statusC | Get durable operation result. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 14 tools
Most tools target distinct resources and lifecycle steps (list vs find metadata, begin_enrollment vs result vs decline, prepare_request vs execute_operation). The main ambiguity is between vault_status and get_connection_status, and the trio vault_list / list_connections / list_operations could be momentarily confused since all are read-only metadata listings. Descriptions mostly disambiguate these, but the 'status' pair overlaps in spirit.
Names are broadly snake_case verb_noun, but split into two families: a vault_-prefixed group and a bare verb group (list_operations, prepare_request, execute_operation). Within the vault family the convention is also mixed, with verbs (vault_list, vault_find, vault_decline) alongside noun-style names (vault_status, vault_result). Readable, but not a single predictable pattern.
Fourteen tools sits comfortably in the sensible 3-15 range and each tool maps to a real step: target selection, enrollment lifecycle, metadata reads, and the operation ticket flow. It is slightly heavy given the enrollment flow spreads over three tools, but nothing feels redundant at scale.
The surface covers connection, discovery, enrollment, operation ticketing, execution, and credential export, but there are notable gaps: no way to update, rotate, revoke, or delete a stored credential or connection. An agent that enrolls a credential has no matching lifecycle tool to retire it, which could cause dead ends.