Connect

Connect

Find meaningful connections between your notes and suggest wikilinks to build a richer knowledge graph.

---

Step 1: Scope

Ask (or infer from context):

AskUserQuestion:

• question: "What scope do you want to search for connections?"
• header: "Connect Scope"
• multiSelect: false
• options:
• label: "specific note", description: "Find connections for a specific note (you'll name it)"
• label: "entire knowledge base", description: "Scan all projects, areas, knowledge, and resources"
• label: "recently added", description: "Notes added in the last 7 days"

---

Step 2: Scan Notes

Use qmd if available for semantic search across notes; fallback: Glob [projects_folder]/**/*.md, [areas_folder]/**/*.md, [knowledge_folder]/**/*.md, [resources_folder]/**/*.md. For each note, extract:

• Title
• Tags (from frontmatter)
• Key concepts mentioned (first 200 words)
• Existing wikilinks already present

If the total note count exceeds 20, delegate to the Knowledge Linker agent instead — see Step 2b.

---

Step 2b: Delegate to Agent (>20 Notes)

If more than 20 notes are in scope: dispatch the Knowledge Linker agent (agents/knowledge-linker.md) as a foreground sub-agent (run_in_background: false, mode: "bypassPermissions"), passing vault_root, knowledge_folder, resources_folder, areas_folder, and projects_folder. The agent always scans the full vault regardless of the scope chosen in Step 1. Proceed to Step 4 immediately (the agent's output replaces Step 3).

For ≤20 notes: continue with Step 3.

---

Step 3: Find Connections

Look for:

| Connection Type | Description | |----------------|-------------| | Conceptual overlap | Both notes discuss the same idea | | Cause-effect | One note explains why something in another happens | | Elaboration | One note goes deeper on a concept mentioned in another | | Contrast | Notes present different perspectives on the same topic | | Sequence | Notes form a natural progression of ideas | | Application | One note shows how to apply a concept from another |

---

Step 4: Present Suggestions

Group suggestions by note. For each suggestion, show:

──────────────────────────────────────────────────────────────
🔗 Connect — {N} suggestions found
──────────────────────────────────────────────────────────────
[{n}/{N}] `{Note A}` → `{Note B}`
  Type: {connection type}
  Reason: {reason}
  Add to `{Note A}`:
    Near "...{nearby text}..." → add [[{Note B}]]

No suggestions:

✅ No connections found — notes are already well-linked.

Maximum 10 suggestions. Ask user to approve each batch before implementing.

If more than 10 suggestions found, show first 10 and add: (showing 10 of {M} — run /connect again to see more)

---

Step 5: Implement Approved Links

For each approved suggestion:

• Read the source note
• Find the suggested location (by nearby text)
• Add the wikilink inline or in a "## Related" section at the bottom

After implementing: ✅ Added [[{Note B}]] link to {Note A}.

---

Step 6: Offer Orphan Report

After connecting, optionally run: > Want me to list notes with no outbound links? These "orphan" notes might be missing connections.

List them if yes, let user decide what to do.

---

Step 7: Offer Typed Relationship Frontmatter

Skip this step if no wikilinks were approved or implemented in Step 5.

After implementing approved wikilinks, use AskUserQuestion to offer typed relationship frontmatter:

> Want me to also add typed relationship properties to the frontmatter of connected notes? This makes connections machine-readable and shows relationship types in Obsidian Graph View. (yes / no)

If user agrees:

• For each connection found, determine the best relationship type: uses, depends_on, contradicts, supersedes, caused_by
• Read the source note's frontmatter
• Add or append to the appropriate property list
• Do not duplicate existing entries (check current frontmatter before writing)

---

Step 8: Batch Retro-tag Mode

If the user asks to retro-tag existing notes (e.g. "update all notes in 01-projects/onebrain/"):

• Glob the target folder for .md files
• For each note: read content + existing frontmatter
• Suggest typed relationships based on existing wikilinks and content
• Present suggestions in batches of 5, ask for approval before writing
• Before writing, check existing frontmatter properties and skip any relationship already present to avoid duplicates
• Write approved typed relationships to frontmatter

---

Step 9: Write Log Entry

Follow ../_shared/audit-log-format.md (canonical frontmatter, append-per-day algorithm, run-section heading, failure mode) with:

• Filename: YYYY-MM-DD-connect.md — one file per day.
• Tags: [audit-log, connect]
• Skill: /connect
• No links applied: if no wikilinks were approved/implemented in Step 5 (and no Step 8 retro-tag changes were written), skip writing — there is nothing to log.

Per-skill body template (canonical ## Run HH:MM heading; metadata in first bullet):

## Run HH:MM

- Scope: knowledge-folder
- Total: N links added across M notes

### Wikilinks added
- `03-knowledge/ai/Foo.md` → added `[[Bar]]`, `[[Baz]]`
- `03-knowledge/dev/Qux.md` → added `[[Foo]]`

---

Progress reporting

This skill is long-running. Emit a 1-line status update after each major step so the user can see progress in real time.

In-session format:

→ [step N/M] 

Examples:

→ [step 1/4] reading target note · finding candidate links...
→ [step 2/4] dispatching knowledge-linker sub-agent...
→ [step 3/4] linker returned 14 candidates · ranking...
→ [step 4/4] writing top suggestions + adding to ## Related...

Rules:

• Emit one line per major step (NOT per sub-step or tool call)
• M = total steps known up front (count them before starting)
• Status lines use → [step N/M] prefix exactly so they're visually distinct from skill output
• Do NOT emit heartbeats for fast operations (< 5 seconds)

---

Known Gotchas

• "Entire knowledge base" scope can be very large. If note count exceeds 20, Step 2b delegates to the Knowledge Linker agent — the inline Steps 3-4 path is only for small scopes. Do not try to hold hundreds of notes in context.

• Avoid suggesting wikilinks that already exist. Before proposing a connection between Note A and Note B, check whether [[Note B]] (or [[Note B|...]]) already appears anywhere in Note A. Suggesting existing links adds noise without value.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: onebrain-ai
• Source: onebrain-ai/onebrain
• License: Apache-2.0
• Homepage: https://onebrain.run

Install and usage instructions live in the source repository linked above.