Related Servers

The tool set is organized into clear families (listing, link-graph, retrieval, audit, PDF/canvas/base), and descriptions consistently state when to use each tool over alternatives (e.g. 'symmetric counterpart to', 'use this instead of'). A few pairs genuinely overlap — obsidian_list_notes vs obsidian_get_recent_edits both surface mtime-ordered note lists, the four retrieval tools (search/hyDE/dataview/context_pack) have adjacent boundaries, and stats vs lint_wiki share orphan/broken-link counts — but careful reading resolves them.

The dominant obsidian_verb_noun pattern is consistent and predictable: list_=enumnerate, get_=single-target, read_=content, query_/validate_/lint_/find_ for actions. Deviations exist — obsidian_frontmatter_get and obsidian_frontmatter_search reverse the pattern to noun_verb, and bare names like obsidian_search and obsidian_stats plus noun_noun names like obsidian_hyde_search and obsidian_context_pack break the mold — but none are chaotic and the obsidian_ prefix anchors everything.

34 tools exceeds the 25-tool threshold for 'too many', and several could be consolidated: four overlapping retrieval tools, three mtime-based listing tools, and partially redundant audit surfaces (stats vs lint_wiki vs get_unresolved_wikilinks). The breadth is defensible for a comprehensive Obsidian analysis server, but the selection burden on an agent is real and the set would benefit from trimming or merging.

Read/analysis coverage is impressively broad — notes, links, tags, frontmatter, hybrid search, PDFs with OCR, canvases, bases, chat threads, and graph analytics. However, descriptions repeatedly reference tools absent from this server (obsidian_create_note, append_to_note, rename_note, frontmatter_set, embeddings_search): obsidian_validate_note_proposal recommends a validate→fix→create workflow that cannot complete here, and lint_wiki suggests fixes via tools that don't exist in this set, creating dead ends.