A personal knowledge management and task automation system with dashboard UI, meeting recording, calendar integration, and AI-powered task execution
Critical gaps across naming, descriptions, and schemas. Most tools lack input parameter schemas entirely (inferred from code, not explicitly registered). Descriptions are present but generic (10-50 chars typical). No output schemas documented. Tool names are inconsistent: some use verb_noun (get_overview, write_dashboard), others are noun-only (inbox_send, finish_meeting_session). Security-sensitive tools (inbox_send, setup_slack_app) expose tokens as parameters rather than using server-side injection. Error handling is absent, no recovery guidance or categorization. Composition is poor: inbox_send and inbox_send_token are tightly coupled; meeting ingestion spans two tools without clear sequencing. This server reads like a collection of internal scripts exposed as tools, not a production-grade MCP implementation.
Browse project files in the Clients directory
Generate lean indexes from large journal/state JSON files for faster access
Finish a meeting recording session, transcode to m4a, and fire watcher for transcription
List or retrieve AI task execution status and results
Fetch briefing card data for the dashboard
Fetch Google Calendar events
Fetch the command queue state
Fetch the dashboard overview data including heartbeat status, freshness indicators, and system health
No input schemas visible for 12 of 16 tools. Tools like get_overview, get_progress, get_briefing, get_command_queue, read_dashboard lack any documented parameter structure. Schemas are inferred from code, not explicitly registered.
Credentials and tokens exposed as tool parameters (inbox_send.token, setup_slack_app.config_token). These will be logged in traces and prompt history. Use server-side secret injection via environment variables or vault.
Descriptions are generic and under 50 characters for most tools (get_progress, get_briefing, get_command_queue, read_dashboard). LLMs cannot determine when to select these tools vs similar ones. Descriptions must explain WHAT, WHEN, and WHY.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 39 | 2026-07-28+ | v2 |
Fetch momentum band progress data for the dashboard
Send an inbox item to Slack with human approval gate
Mint a single-use token for sending an inbox item
Index work documents and upload to Google Sheets with Drive links
Receive and store a chunk of meeting audio from the phone
Read the Dashboard.md file content
Create or update a Slack app using manifest configuration
Write content to the Dashboard.md file
No output schemas documented. Tools return data but LLMs cannot plan downstream calls without knowing response structure. Document what fields each tool returns and their types.
No error handling or recovery guidance. Tools fail silently or return raw errors. Implement error categorization (retryable, user-fixable, fatal) and actionable messages like 'File not found. Check path with browse_files().'