MCP Server for Dokploy API - provides tools to interact with Dokploy deployment platform
17 tools with auto-generated schemas from OpenAPI spec. All tools have input schemas (JSON Schema with types), but descriptions are minimal (mostly 'POST /path' or 'GET /path', under 20 chars). No parameter descriptions visible in tool definitions. Tool names follow verb_noun pattern (e.g., 'application-create', 'admin-setupMonitoring') but lack clarity on when to use each variant. Schemas are well-formed with constraints (enums, patterns, minLength), but output schemas are not documented. Error handling and recovery guidance absent. No evidence of idempotency hints or confirmation patterns for destructive operations.
POST /admin.setupMonitoring
POST /application.create
POST /application.delete
POST /application.deployNginxQuickstart
GET /application.one
POST /application.redeploy
POST /application.reload
Tool descriptions are minimal (5-10 chars, e.g., 'POST /application.create'). LLMs cannot determine when to select a tool or what it does without a proper description. Rubric requires 10-1024 chars with context on WHAT, WHEN, and WHY.
No parameter descriptions visible in tool definitions. Each parameter (e.g., 'metricsConfig', 'appName', 'environmentId') lacks a description explaining its purpose, valid values, or constraints. Rubric requires every parameter to have a description.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 51 | <=2025-11-25 | v2 |
POST /application.saveBitbucketProvider
POST /application.saveBuildType
POST /application.saveDockerProvider
POST /application.saveEnvironment
POST /application.saveGitProvider
POST /application.saveGiteaProvider
POST /application.saveGithubProvider
POST /application.saveGitlabProvider
POST /application.start
POST /application.stop
Output schemas are not documented. Tool definitions show input schemas but no description of what fields are returned. LLMs cannot plan downstream tool calls or extract required data without knowing the response structure.
Destructive operations (application-delete) lack confirmation or dry-run patterns. No evidence of error handling guidance or recovery steps. Agents may delete applications without safeguards.
Tool naming uses hyphenated format (e.g., 'application-create', 'admin-setupMonitoring') which is non-standard. Rubric baseline shows 90% of A+ tools use verb_noun with underscores (e.g., 'create_application'). Hyphenation may confuse LLM parsing.