LastPass MCP Server provides vault management tools via the Model Context Protocol, enabling AI assistants to manage LastPass vault entries remotely using the standard MCP protocol over HTTP Streamable transport.
LastPass MCP has 7 well-named tools with clear verb-noun patterns (lastpass_login, lastpass_search, etc.). All tools have descriptions and input schemas with parameter types. However, descriptions lack depth, most are 60-90 chars and omit critical details like prerequisites, error guidance, and chaining information. Output schemas are not documented. Parameter descriptions are minimal and lack format/constraint details. Error handling is not evident in the source. Missing field naming clarity between tools, e.g., lastpass_show returns unspecified fields, forcing LLMs to guess what data is available. Security concern: password parameter in lastpass_login is exposed as a plaintext parameter (violation of secret-injection pattern).
Create a new vault entry. Specify type as 'password' or 'paymentcard' and provide the relevant fields.
Delete a vault entry permanently by ID.
Authenticate to LastPass with email and master password. Creates or refreshes the LastPass session associated with the current Bearer token.
Terminate the current LastPass session and invalidate the Bearer token.
Search vault entries by regular expression pattern. Matches against name, url, and username (case insensitive). Optionally filter by entry type.
Show full details of a vault entry by ID. Returns all fields including password or payment card details.
Password parameter exposed as plaintext tool input in lastpass_login. Violates secret-injection pattern, credentials must be server-side injected, not passed as parameters. Agent traces will log plaintext passwords.
Output schemas are not documented. LLMs cannot determine what fields lastpass_show, lastpass_search, lastpass_create, and lastpass_update return. Forces LLMs to guess field names for downstream chaining.
Destructive tool (lastpass_delete) lacks confirmation or dry-run capability. No guidance for recovery or undo. Agents can permanently delete vault entries without safeguards.
Parameter constraints are not formalized. 'type' parameter in lastpass_search and lastpass_create should be enums ('password' | 'paymentcard'), not free-form strings. Allows LLMs to pass invalid values.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | C | 60 | <=2025-11-25 | v2 |
Update an existing vault entry. Specify the entry ID and the fields to update.
Parameter descriptions lack depth. None mention format constraints, validation rules, required vs optional fields, or default behavior. E.g., 'pattern' in lastpass_search says 'regex pattern' but does not explain case sensitivity, escape rules, or performance limits.
No error handling guidance. Tool descriptions do not explain what happens on failure (auth failure, entry not found, permission denied, etc.) or how to recover. LLMs receive no actionable error messages.
lastpass_search lacks pagination parameters and result limits. No mention of how many items are returned, whether results are paginated, or total count. Large vaults could return thousands of entries, exhausting context.
Tool composition issue: lastpass_login and lastpass_logout should explain session lifetime and whether multiple simultaneous sessions are allowed. No guidance on Bearer token lifecycle or how to handle expired sessions.