A product development environment for the AI era with MCP server capabilities for project management, backlog tracking, and agent integration
Vogt exposes 61 tools via STDIO with minimal schema visibility. Tool names follow verb_noun convention (vogt_backlog, vogt_work_create, etc.), which is positive. However, critical gaps undermine quality: (1) No input schemas are visible in the provided source code, only tool names and descriptions appear in stdio.py. (2) Descriptions are extremely terse (10-30 chars), far below the 50-200 char baseline for LLM-optimized tools. E.g., 'List backlog items' (19 chars) lacks context on when to call it vs vogt_work_list. (3) No parameter documentation is evident, tools like vogt_work_create and vogt_work_update have no visible input specs. (4) Output schemas are undocumented. (5) Error handling is not visible. Per the hard rules: tools with no visible input schema score 0 on schema; descriptions under 20 chars score 0-20 on description. Average across 61 tools: ~15/100.
List actors
List agent CLI tools
List audit entries
List backlog items
List boards
List bug items
Check compliance status
Adopt contract
No input schemas visible for any of 61 tools. This blocks LLM parameter inference and validation.
All tool descriptions are under 20 characters (e.g., 'List backlog items', 'Get project details'). Descriptions lack context on when to call each tool, dependencies, and return structure. LLMs cannot distinguish vogt_work_list from vogt_backlog without richer descriptions.
No parameter documentation visible. Tools like vogt_work_create, vogt_work_update, vogt_project_update have no visible parameter descriptions, types, or constraints. LLMs cannot infer what fields are required, what values are valid, or what format is expected.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 35 | 2026-07-28+ | v2 |
Mark contract as applicable
Check contract
Decline contract
Mark contract as inapplicable
Get coverage information
Get dependency information
Detect drift
List drift items
Resolve drift
List events
Get forge account status
List forge actions
Onboard to forge
List forge repositories
Archive inbox item
List inbox items
Restore inbox item
Snooze inbox item
Create initiative
List initiatives
Publish initiative
Create label
List labels
List observations
Get place metrics
Get project brief
Get project details
List projects
Transition project state
Update project
List session history
List sessions
Get session log tail
Search session output
Start a session
Stop session
Get status information
Suppress item
List suppressions
Revoke suppression
Perform sweep operation
Get context and reasoning
Adopt work item
Bind work item to branch
Add comment to work item
Create work item
Get work item details
List work items
Create relationship between work items
Transition work item state
Remove relationship between work items
Update work item
List workflows
No output schemas documented. LLMs cannot plan downstream tool calls or extract required fields (e.g., work_id from vogt_work_create to pass to vogt_work_comment). Pagination, field structure, and return types are invisible.
No error handling guidance visible. Tools like vogt_work_create and vogt_contract_adopt (WRITE operations) have no visible error recovery hints, retryability classification, or actionable error messages. Agents cannot self-correct on failure.