MCP server for agent task workflows with phases, prompts, gates, and multi-agent coordination
Three tools with clear naming (verb_noun pattern) and reasonable descriptions (100-200 chars). All have input schemas with typed parameters and descriptions. However, output schemas are not documented in the visible code, parameter constraints lack enums/ranges, and error handling guidance is absent. No tool annotations (readOnlyHint/destructiveHint). The get_prompts tool is well-described but lacks output schema documentation. add_overlay and remove_overlay lack detail on what 'overlay diff' and 'active overlay list' contain. No validation rules or recovery guidance in descriptions.
Add an overlay to a worker's active overlays, persisting the change to the database and returning overlay diff and active overlay list
Get workflow prompts and guidance. Without parameters: lists all prompt triggers. With status/phase: returns the prompts that would fire for entering that state. Useful for understanding what guidance applies to a transition before making it.
Remove an overlay from a worker's active overlays, persisting the change to the database and returning overlay diff and active overlay list
Output schemas not documented. get_prompts returns JSON with 'triggers', 'status', 'phase', 'prompts', 'count' fields, but LLMs cannot infer structure without explicit schema. add_overlay and remove_overlay return 'overlay diff' and 'active overlay list', structure undefined.
No tool annotations. add_overlay and remove_overlay are destructive (WRITE risk) but lack destructiveHint annotation. get_prompts is read-only but lacks readOnlyHint. Agents cannot infer safety properties.
Parameter descriptions lack validation rules. 'worker_id' and 'overlay' accept free-form strings with no format, length, or allowed-value constraints. No guidance on what constitutes a valid overlay name.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 47 | <=2025-11-25 | v2 |
Error handling absent from descriptions. No guidance on what happens if worker_id does not exist, overlay is invalid, or database write fails. Agents cannot plan recovery.
add_overlay and remove_overlay descriptions are vague. 'Add an overlay to a worker's active overlays' does not explain what an overlay is, when to use it, or what side effects occur. Agents cannot determine if this is the right tool.