MCP server exposing intent-shaped tools over the Obsidian second-brain vault
Two tools with strong naming (vault_status, vault_list) and comprehensive descriptions (200+ chars each). Both follow verb_noun pattern. However, input schemas are not visible in the provided source excerpt, only handler signatures shown. vault_list description is truncated mid-sentence ('last_entry_date is the latest `### YYYY-MM-DD` heading or dat'). No parameter descriptions visible for either tool. Error handling via ToolError class exists but recovery guidance is minimal. No output schema documentation provided. Tool composition is sound (single responsibility), but schema completeness cannot be verified from the code sample.
List notes, filtered by frontmatter type, top-level folder, frontmatter status, linked goal, modification date, or last dated entry. Template notes are excluded unless include_templates=true. Set latest=true to get only the newest date-named note in a folder (use this instead of listing a folder and eyeballing the max filename). last_entry_date is the latest `### YYYY-MM-DD` heading or dat
Vault orientation in one call: today's local date, git dirty state, note counts by type, latest synthesis, daily note, and review week, unresolved/orphan link counts, task counts, and notes changed in the last 24 hours. Call this first in any standup, weekly review, or nightly run instead of exploring the vault by hand. git_error is non-null when git is enabled but unusable — snapshots will fail until it is fixed.
Input schemas not visible in source code. Both tools show empty or missing schema definitions in the provided excerpt. Cannot verify parameter types, constraints, or descriptions.
vault_list description is truncated mid-sentence ('last_entry_date is the latest `### YYYY-MM-DD` heading or dat'). Incomplete descriptions prevent LLMs from understanding tool purpose and when to invoke it.
No parameter descriptions visible. Even if schemas exist in full source, LLMs cannot infer meaning from parameter names alone (e.g., 'type', 'status', 'limit' are ambiguous without context).
No output schema documentation. LLMs cannot plan downstream tool calls or extract required fields without knowing the response structure.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 47 | 2026-07-28+ | v2 |
Error handling via ToolError exists but lacks recovery guidance. Errors should tell LLMs what to do next (e.g., 'git is disabled, enable it in config or retry without git checks').