/ingest

description: "Ingest source material into an active wiki. Accepts URLs, file paths, PDFs, freeform text, or processes the inbox. Supports tweets via Grok MCP."
argument-hint: "<url|filepath|\"text\"> [--type articles|papers|repos|notes|data] [--title \"Title\"] [--inbox] [--keep] [--wiki <name>] [--local] [--auto-classify] [--new-topic <name>] [--project <slug>] [--include-archived]"
allowed-tools: Read, Write, Edit, Glob, Grep, Bash(ls:*), Bash(wc:*), Bash(date:*), Bash(mv:*), Bash(mkdir:*), Bash(basename:*), Bash(file:*), Bash(curl:*), Bash(mktemp:*), Bash(rm:*), Bash(pdftotext:*), Bash(python3:*), WebFetch, WebSearch

Your task

**Resolve the wiki.** Do NOT search the filesystem or read reference files — follow these steps: 1. Read `$HOME/.config/llm-wiki/config.json`. If it has `hub_path`, expand leading `~` only (not tildes in `com~apple~CloudDocs`) and prefer that path; use `resolved_path` only as a fallback cache when the expanded `hub_path` is unavailable and `resolved_path` is initialized. If config has only `resolved_path`, use it. If the configured path can be statted but reading `wikis.json` or listing `topics/` fails with `Operation not permitted`, stop and ask the user to grant Full Disk Access/iCloud Drive access to the launcher; do not fall back to `~/wiki` or `resolved_path`. Do not write machine-specific `resolved_path` into shared configs. 2. If no config → read `$HOME/wiki/_index.md`. If it exists → HUB = `$HOME/wiki`. If nothing found, ask the user where to create the wiki. 3. **Wiki location** (first match): `--local` → `.wiki/` in CWD; `--wiki <name>` → `HUB/wikis.json` lookup with portable path resolution (`<HUB>`, `~`, absolute, or HUB-relative); if the registry path is stale, fall back to `HUB/topics/<name>`; CWD has `.wiki/` → use it; else → HUB. 4. Read `<wiki>/_index.md` to verify. If missing and `--new-topic <name>` is set → create the topic wiki (see below). If missing and no `--new-topic` → stop with "No wiki found. Use `--new-topic <name>` to create one, or run `/wiki init` first."

Read the ingestion protocol at `skills/wiki-manager/references/ingestion.md` and the structure spec at `skills/wiki-manager/references/wiki-structure.md`. Then ingest source material.

Collection handoff

If the user is trying to import a bounded upstream corpus rather than one source item, stop and hand off to `/wiki:ingest-collection` with the same wiki flags. Signals include: "import wiki", "mirror wiki", "bulk ingest", "ingest collection", "ingest repo", "import repo", "split CSV into messages", "Wayback snapshots", MediaWiki dumps (`.xml`, `.xml.bz2`, `.xml.gz`), MediaWiki `api.php`, CDX API URLs, or a GitHub/GitLab repo plus words like "all", "docs", "BIPs", "wiki", or "collection".

Do not reduce a collection to the repository README or one fetched web page. Collection imports need a manifest plus one raw child source per upstream page/spec/proposal.

Inventory and dataset awareness

If the user asks to "track", "watch", "keep inventory", "candidate", "queue", or "decide later" instead of ingesting now, stop and hand off to `/wiki:inventory`. A durable candidate record is better than ingesting a source the user has not accepted.

If the source is a large dataset, archive, compressed dump, database, or query-oriented corpus, stop and hand off to `/wiki:dataset` or `/wiki:ingest-collection` as appropriate. Do not squeeze large data into `raw/data/` because it is technically readable.

After a successful ingest, check active inventory records by title/source when cheap. If a matching record exists, report the suggested linkage and status change (`sources:` add raw path, maybe status `ingested`) instead of silently leaving the tracking state stale.

Archive awareness

Do not ingest into archived topic wikis by default. If `--wiki <name>` resolves to `status: archived` or a path under `topics/.archive/`, stop and ask the user to restore it with `/wiki:archive restore <name>` or rerun with `--include-archived`. If the user explicitly includes archived content, write only inside that archived topic path and keep the topic archived.

`--new-topic` branch

When `--new-topic` is set, override the standard resolution: 1. Derive a slug from the topic name: lowercase, hyphens, no special chars, max 40 chars 2. If HUB doesn't exist, create it (wikis.json + _index.md + log.md + topics/) 3. Create the new topic wiki at `HUB/topics/<slug>/` following the full init protocol (directory structure, .obsidian/, empty _index.md files, config.md, log.md) 4. Register in `HUB/wikis.json` using a portable relative path (`topics/<slug>`) and update hub `_index.md` 5. Target this new wiki for ingestion

`--project <slug>` flag

**--project <slug>**: Tag the ingested source with `project: <slug>` frontmatter. The raw file still lands in the normal `raw/<type>/` location (sources are shared across projects by design). The project tagging propagates later when sources are compiled into wiki articles. See `references/projects.md`.

If `--project <slug>` is set, verify the project exists at `<wiki-root>/output/projects/<slug>/WHY.md`. If not, fail early: `Project "<slug>" does not exist. Create it first with /wiki:project new <slug> "goal".`

There is no ambient project focus — pass `--project` explicitly when you want project scope. The focus-session mechanism (`.wiki-session.json`) was removed in the v0.2 projects simplification; see `references/projects.md` § "Focus" for the rationale.

Parse $ARGUMENTS

• **--new-topic <name>**: Create a new topic wiki and ingest the source into it. Works from any directory. Derives the wiki slug from the name (e.g., "knives" → `knives`, "machine learning" → `machine-learning`).
• **--inbox**: Process all files in the wiki's `inbox/` directory
• **--keep**: When processing inbox, keep originals (move to .processed/ instead of deleting)
• **--type**: Force source type (articles, papers, repos, notes, data). Default: auto-detect.
• *