wiki-scan

wiki-scan

Read skills/wiki-schema/SKILL.md fully before proceeding.

Invoke with: /wiki-scan [directory]

If no directory argument is given, scan all Layer 1 directories plus .raw/.

---

Step 1 — Resolve VAULT_PATH and target directories

Vault path: ~/obsidian_vault by default; set VAULT_PATH to override. WSL users with a Windows-side vault must set VAULT_PATH explicitly (e.g. /mnt/c/Users/<name>/obsidian_vault).

Default scan targets (all Layer 1 + inbox):

Meetings/
Daily/
Projects/
Customers/
People/
Clippings/
Research/
.raw/

If a directory argument was given, scan only that directory tree.

---

Step 2 — Read _manifest.json

Read wiki/_manifest.json. If it does not exist, treat all files as new.

---

Step 3 — Enumerate files and compute hashes

For each .md file in the target directory tree:

find "<dir>" -name "*.md" -not -path "*/_processed/*"

For each file, compute hash:

md5sum "<filepath>" | cut -d' ' -f1

Classify each file:

• New: path not in manifest → needs ingest
• Changed: path in manifest, hash differs → needs re-ingest
• Unchanged: path in manifest, hash matches → skip

---

Step 4 — Report and confirm

Show the scan summary before processing anything:

Wiki Scan Results — <directory or "all sources">
─────────────────────────────────────────────────
New files (N):
  - Meetings/ACME/2026-04-15 Sprint Review.md
  - .raw/articles/mcp-patterns-2026-04-20.md

Changed files (M):
  - Projects/Products/P1234 MCP server for Acme.md  (hash changed)

Unchanged files: K (skipping)

Proceed with ingesting N new + M changed files? [yes/no]

Wait for explicit confirmation before proceeding. If the user says no, stop.

---

Step 5 — Ingest each new/changed file

Process files sequentially. For each file, follow the full wiki-ingest workflow (Steps 2–10 of wiki-ingest skill — delta check, read, extract, resolve, write pages, update index, log, manifest, hot cache).

Update _log.md once per file during processing (not a single batch entry at the end).