MCP server for Beacon, a visual planning surface for coding agents. Provides tools for feature planning, task management, and workspace interaction.
Single tool 'beacon_feature' has a visible schema with 13 parameters, but critical gaps undermine quality. Description (127 chars) is present but vague, it lists actions without explaining WHEN to use each or what the tool returns. Parameters lack individual descriptions entirely (0/13 have descriptions per the schema). The schema shows types and enums for 'action' and 'status', but most parameters (title, id, category, priority, front, kind, layer, description, parentId, parentTitle, items, features, files, architecture) have no descriptions, violating the baseline that 100% of A+ tool params have descriptions. No output schema is documented. Error handling is absent from the visible code. The tool conflates multiple concerns (add, start, subtasks, done) into one dispatcher, violating single-responsibility. No evidence of idempotency, pagination, or recovery guidance.
Dispatcher for feature lifecycle management: add (create a card in BACKLOG/PENDING), start (mark IN_PROGRESS), subtasks (add child tasks), done (complete features and register files/architecture)
No parameter descriptions. All 13 parameters lack descriptions (title, id, category, priority, front, kind, layer, status, description, parentId, parentTitle, items, features, files, architecture). Baseline: 100% of A+ tools have param descriptions.
Tool description is vague (127 chars). States actions but not WHEN to use each, what it returns, or prerequisites. Baseline: 10 - 1024 chars, but must answer What/When/Return. Current text lists actions without context.
No output schema documented. LLMs cannot plan downstream calls or extract data without knowing response structure. Baseline: 100% of A+ tools have documented return types.
Tool combines multiple concerns (add, start, subtasks, done) into one dispatcher. Violates single-responsibility pattern. Should split into separate tools: create_feature, start_feature, add_subtasks, complete_feature.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 38 | 2026-07-28+ | v2 |
No error handling or recovery guidance visible. No indication of retryability, user-fixable errors, or actionable error messages. Baseline: errors must guide LLM on next steps.