Zero-trust credential gateway for AI Agents — proxy-only, passwords never enter LLM context. Provides READ-ONLY access to password entries via MCP protocol.
Two tools with minimal documentation. list_passwords has a basic description (READ-ONLY access) but no visible input schema. get_password has a description and visible input schema with two optional parameters (domain, name), but both lack detailed descriptions explaining format/constraints. No output schemas documented. No error handling guidance. No parameter validation rules stated. Descriptions are under 100 chars, below the 194-char baseline for production tools. Tool composition is reasonable (two distinct read-only operations), but schema completeness and parameter documentation are weak.
Get a specific password entry by domain or name. Returns proxy configuration (alias, proxyUrl, upstreamUrl) instead of exposing the actual password.
List all password entries with metadata (READ-ONLY access)
list_passwords has no visible input schema in source code. Cannot verify parameter definitions, types, or constraints.
get_password input parameters (domain, name) lack detailed descriptions. No format constraints, mutual exclusivity rules, or usage guidance provided. LLM cannot infer when to use domain vs name.
No output schemas documented for either tool. LLMs cannot plan downstream operations or extract required fields. Proxy configuration response structure (alias, proxyUrl, upstreamUrl) mentioned in description but not formalized.
No error handling guidance. Tools do not document what happens on invalid domain, missing entry, or vault access failure. LLM has no recovery path.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 42 | 2026-07-28+ | v2 |
Tool descriptions are 50 - 100 chars, well below 194-char baseline. Lack context on WHEN to use each tool, prerequisites, and expected return structure.