// Analyze and build knowledge graph links in Obsidian vault. Runs a deterministic script for analysis, then the agent adds semantic links to orphan files. Three domains: Personal, Business, Projects.
Todoist integration via mcp-cli for task management
Vault health monitoring, MOC generation, link repair, and system evolution. Use this skill when checking vault quality metrics, regenerating MOC indexes, fixing broken links, finding backlinks, or triggering weekly system reflections. Also use it when the user mentions health score, link density, orphans, dead-ends, or description coverage.
Personal assistant for processing daily voice/text entries from Telegram. Classifies content, creates Todoist tasks aligned with goals, saves thoughts to Obsidian with wiki-links, generates HTML reports. Triggers on /process command or daily 21:00 cron.
| name | graph-builder |
| description | Analyze and build knowledge graph links in Obsidian vault. Runs a deterministic script for analysis, then the agent adds semantic links to orphan files. Three domains: Personal, Business, Projects. |
Analyze vault link structure and build meaningful connections between notes.
+------------------------------------------+
| 1. SCRIPT analyze.py (deterministic) |
| -> vault-graph.json + report.md |
+-----------------+------------------------+
|
v
+------------------------------------------+
| 2. AGENT (you) |
| -> orphan analysis |
| -> semantic links |
| -> frontmatter updates |
| -> HTML report |
+------------------------------------------+
/graph -- manual rundbrain-processor in chain (optional)uv run vault/.claude/skills/graph-builder/scripts/analyze.py
Script does:
[[path]] and [[path|alias]]vault/.graph/vault-graph.jsonvault/.graph/report.mdCRITICAL: Always run the script first! It provides up-to-date data.
Read vault/.graph/vault-graph.json:
{
"stats": {
"total_files": 331,
"orphan_files": 147,
"broken_links": 167
},
"orphans": ["path/to/file.md", ...],
"broken_links": [{"from": "...", "to": "..."}, ...]
}
For each orphan file (start with non-daily files):
Read the file content
Find potential links:
business/crm/{company}.mdSuggest links -- add to frontmatter related: []
| Mention | Link |
|---|---|
| Acme Corp, Acme | [[business/crm/acme-corp]] |
| PhoneBrand | [[business/crm/phone-brand]] |
| TechCo | [[business/crm/tech-co]] |
| SchoolCo | [[projects/leads/school-co]] |
| AutoDealer | [[projects/leads/auto-dealer]] |
| AI, Claude, GPT | thoughts/learnings/ with AI topic |
Link format in frontmatter:
---
type: learning
domain: personal
tags: [ai, marketing]
related:
- "[[business/crm/acme-corp]]"
- "[[projects/leads/school-co]]"
- "[[thoughts/learnings/agent-memory-system]]"
---
Rules:
"[[path]]" format with quotesCheck MOC files:
MOC/MOC-learnings.md -- add new learningsMOC/MOC-projects.md -- create if missingReturn RAW HTML for Telegram:
Graph Builder Report
Analyzed:
Business: {files} files, {links} links Projects: {files} files, {links} links Personal: {files} files, {links} links
Links added: {count}
Needs attention:
Details: vault/.graph/report.md
thoughts/ -- learnings, ideas, reflectionsgoals/ -- yearly, monthly, weeklydaily/ -- raw entriesMEMORY.mdbusiness/crm/ -- clients + deals (N files)business/network/ -- org structurebusiness/_index.mdprojects/clients/ -- project clientsprojects/leads/ -- leadsprojects/_index.mdreferences/domains.md -- domain detailsreferences/entities.md -- entity patternsreferences/frontmatter.md -- frontmatter schema