| name | wiki-ingest-triage |
| description | Triage a URL, PDF, or file for ingestion into the knowledge wiki — assess fit, pick domain, run ingest |
| triggers | ["wiki-ingest-triage","wiki ingest-triage"] |
What This Does
Evaluates whether a source belongs in the wiki, picks the right domain, and
runs wiki ingest. Covers the full triage loop: fit check → domain selection →
ingest → confirm frontmatter.
Reference conventions: WIKI.md § 3 (source types), § 9 (hard rules), § 10 (filter policy).
Usage
/wiki-ingest-triage <url-or-path> # Triage a single source
/wiki-ingest-triage <url> --domain glp1 # Skip domain selection, use glp1
Execution Steps
-
Identify source type: Determine type from URL/path extension per WIKI.md § 3.1.
Supported: web, youtube, arxiv, pubmed, pdf, voice, note, image, etc.
-
Assess fit: Ask one question — does this source contribute novel, citable
evidence to any existing domain? If no, say so and stop.
-
Select domain: Run wiki status to see active domains. Match source topic
to the best-fit domain slug. If ambiguous, ask the user.
-
Plan before ingest (hard rule — WIKI.md § 9): State which pages will be
created before running any command.
-
Run ingest:
.venv/bin/wiki ingest <path-or-url> --domain <slug>
Add --draft if the source needs citation review before finalization.
-
Confirm: Read the created raw page and wiki/sources page. Verify frontmatter
has domains:, content_hash:, ingested_at:. Report the source ID.
Key constraints
- Never write directly to
wiki/ or raw/. All writes go through the gateway (WIKI.md § 9).
- Large batches (50+ sources) belong in
wiki batch-ingest, not this skill.
- If the source type is unsupported, report the gap — don't attempt a manual write.