Sw Mcp Vault Knowledge — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited Sw Mcp Vault Knowledge (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
A read-only MCP server that turns an Obsidian vault's typed knowledge graph into three tools any MCP client can call. Most knowledge tools can search. Few can tell you where your own notes disagree with each other. This one traverses a six-relation reasoning graph and surfaces the sharpest edge — contradicts. Local-first, stdio, no auth, no cloud, no native build.
The full reasoning lives in docs/EXPLANATION.md.
| Tool | Returns | |
|---|---|---|
search_concepts(query, limit=5) | Semantically ranked concept/connection articles. | |
| `find_contradictions(scope='all'\ | 'recent_30d')` | Active contradicts edges from the reasoning graph. |
get_article(slug) | One article with frontmatter, body, and resolved wikilinks. |
Ships with a synthetic espresso vault, so it works the moment it installs.
{
"mcpServers": {
"vault-knowledge": { "command": "npx", "args": ["-y", "@swins/vault-knowledge-mcp"] }
}
}Then ask: "find contradictions in the vault."
{
"mcpServers": {
"vault-knowledge": {
"command": "npx",
"args": ["-y", "@swins/vault-knowledge-mcp"],
"env": {
"VAULT_DB": "/abs/path/.vault-index.db",
"VAULT_ROOT": "/abs/path/vault"
}
}
}
}Only knowledge/{concepts,connections,qa} is ever readable. Scope is config-only — never a request parameter, so a client can't widen it.
Node ≥ 22.5 (built-in node:sqlite). search_concepts downloads the nomic-embed-text-v1.5 model once on first run.
MIT.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.