What does this skill do, and when should you use it?
wiki-query is the query-focused skill in the claude-obsidian collection. It answers questions from an Obsidian wiki vault by reading the hot cache, master index, and relevant pages in a deliberate order. It can use the optional wiki-retrieve pipeline to rank candidate pages, then synthesizes answers with page-level or chunk-level citations. Strong answers can be filed back into the vault as question pages, making it a fit for users who already maintain—or are willing to build—a Markdown knowledge base.
Selects Quick, Standard, or Deep query mode; reads wiki/hot.md and wiki/index.md; opens relevant wiki pages for Standard and Deep queries and may follow wikilinks to depth two; when wiki-retrieve is provisioned, runs python3 scripts/retrieve.py "<the user's question verbatim>" --top 5, reads the returned candidates, and synthesizes with chunk-level citations; reads through Obsidian CLI, MCP, or filesystem transport; cites specific wiki pages; and files suitable answers under wiki/questions/ while updating wiki/index.md and wiki/log.md.
- A researcher wants a cited answer from an existing Obsidian research vault.\n
- A project team member needs to query an entity, concept, or source in a Markdown-based project wiki.\n
- A personal knowledge-management user wants quick answers from recent context without rereading the full vault.\n
- A user needs a cross-page synthesis and chooses Standard or Deep mode for broader retrieval.\n
How do you install this skill?
- The instruction to file answers back to the wiki conflicts with allowed-tools: Read, Glob, and Grep; it should either be limited to suggesting a save or explicitly require write permission.
- Deep mode mentions optional web supplementation without declaring web-tool availability, consent flow, or external data-flow boundaries; handling of sensitive vault content is unclear.
- No tests, representative outputs, or failure-case evidence are supplied for wiki-query's key paths.
- The publisher is not verified by the FollowSkills enterprise registry and should be treated as unknown; the license is MIT, but maintenance ownership and update path are not sufficiently explicit.
- Shell / CLI
- Network access
- Local filesystem
- MCP Server
Obsidian vaultobsidian-cli or filesystem accessPython 3.10+ for optional retrievalmcp-obsidian or mcpvault for MCP transport
Install the full repository with git clone https://github.com/AgriciDaniel/claude-obsidian, then run cd claude-obsidian && bash bin/setup-vault.sh. Keep the skill at skills/wiki-query/SKILL.md and open the repository folder as an Obsidian vault. The README does not document a separate installation command for wiki-query alone.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/AgriciDaniel/claude-obsidian.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/wiki-query" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
In Claude Code, use a supported natural-language trigger such as what do you know about X?, find in wiki: X, or query deep: compare A and B. Use query quick: ... for a cache-and-index lookup; omit the mode for Standard; and use query deep: ... for comprehensive synthesis. If the wiki cannot answer the question, the skill should identify the specific gap instead of fabricating an answer from training data.
What are this skill's strengths and limitations?
- Uses a hot-cache → index → relevant-pages retrieval order to limit unnecessary reading and token use.
- Produces answers with traceable citations to specific wiki pages.
- Offers Quick, Standard, and Deep modes for different query depths.
- Can file valuable answers back into the vault so knowledge compounds over time.
- Requires useful content in wiki/hot.md, wiki/index.md, and relevant pages; it cannot reliably answer from an empty or sparse vault.
- Standard and Deep behavior depends on the vault structure and an available read transport.
- The SKILL.md does not fully document setup procedures for MCP, Obsidian CLI, or optional retrieval components.
- No standalone test suite or verification evidence for other platforms is provided in the skill source.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Obsidian Wiki Query this page | 50 · Use with care | ★ 16k | 1mo ago | MIT |
| Wiki Log Folder | 55 · Use with care | ★ 16k | 1mo ago | MIT |
| Conversation Archivist | 49 · Use with care | ★ 16k | 1mo ago | MIT |
| Wiki Knowledge Companion | 49 · Use with care | ★ 16k | 1mo ago | MIT |
| Wiki Ingest for Obsidian | 49 · Use with care | ★ 16k | 1mo ago | MIT |
How did FollowSkills review this skill?
The skill restricts tools to Read, Glob, and Grep, uses a least-read hot→index→page sequence, and requires wiki-page citations without fabrication. It does not clearly specify sensitive-data handling, external data flows, user confirmation, or rollback, and its instruction to file answers conflicts with the declared tool permissions, so points are deducted.
Query modes, read order, retrieval fallback, and gap handling are specified coherently. However, behavior depends on transport configuration, optional scripts, and referenced skills; abnormal-input handling and diagnosable failure feedback are limited. The static cap keeps this below 10.
The audience, Quick/Standard/Deep scenarios, trigger phrases, and wiki-coverage boundary are reasonably clear. Language support, varied vault layouts, CLI/MCP unavailability, and the Deep mode's web capability are insufficiently bounded, so points are deducted.
Frontmatter, mode guidance, token budgets, workflows, and citation format are well organized. The repository context supplies MIT licensing, version information, installation guidance, and a security policy, but the selected skill lacks its own changelog, explicit maintenance owner/update path, and troubleshooting guidance, so it is not full marks.
The core task—synthesizing cited answers from hot cache, index, and relevant pages—has direct practical value. However, filing answers cannot be performed within the declared read-only tool set, and no representative wiki-query outputs are verified; the static cap keeps this at or below 7.
The skill gives auditable read-order, citation, fallback, and no-fabrication rules. Repository CI and tests exist, but the supplied tests mainly cover other scripts and do not verify wiki-query's key paths. Independent reproduction evidence is therefore thin, warranting only limited credit.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →