/ingest
Ingest a source (paper, article, transcript, PDF, notes) into the LLM Wiki.
$ npx -y skills add praneybehl/llm-wiki-plugin --agent claude-codeHow it fires
How this command gets triggered: by you, by Claude, or both.
- Fires itselfClaude auto-loads it when your prompt matches the work.
- You can call itInvoke it directly when you want it.
- Slash command
/ingest
Context preview
What this command does when you run it.
Ingest a source (paper, article, transcript, PDF, notes) into the LLM Wiki.
Command definition
ingest.mddescription: Ingest a source (paper, article, transcript, PDF, notes) into the LLM Wiki.
argument-hint: "<source-path-or-url-or-description>"
Ingest the source into the wiki using the `llm-wiki` skill's ingest workflow.
Source: $ARGUMENTS
Follow the full ingest procedure documented in the skill:
1. Read `wiki/SCHEMA.md` first if you haven't this session — it may override default conventions. 2. Place the raw source in `raw/` if it isn't already there. Use a slugified filename. 3. Read the source. Chunk-read if it's large (over ~5000 words / a long PDF) — never load the whole thing into context if it would consume more than ~25% of the context window. 4. Briefly discuss the key takeaways with me before writing anything to the wiki — what stands out, what connects to existing pages, what's surprising. 5. Survey the wiki to identify which existing pages this source touches; read each candidate to confirm. 6. Write the source-summary page in `wiki/sources/`, surgically update touched entity/concept pages with `str_replace`, create new pages for new entities/concepts (each with at least one inbound link), update the index, append one line to `log.md`. 7. If `wiki/graph/ontology.yaml` exists and the ingest added or could add typed `graph.relationships[]`: add typed edges only when the source explicitly supports them (predicate, source-page slug, evidence quote, confidence, status). Then run `uv run --script "${CLAUDE_PLUGIN_ROOT}/skills/llm-wiki/scripts/wiki_graph_lint.py" "${PWD}/wiki"`, triage findings with me, and run `uv run --script "${CLAUDE_PLUGIN_ROOT}/skills/llm-wiki/scripts/wiki_graph_extract.py" "${PWD}/wiki"`. Append a ` graph: +N nodes, +M typed edges` sub-line under the ingest entry in `log.md`. 8. Tell me what you did when you're done: pages touched, pages created, contradictions flagged, follow-ups worth investigating.
If the wiki doesn't exist yet, suggest running `/wiki:init` first.
Read more
description: Ingest a source (paper, article, transcript, PDF, notes) into the LLM Wiki. argument-hint: "<source-path-or-url-or-description>"
Ingest the source into the wiki using the `llm-wiki` skill's ingest workflow.
Source: $ARGUMENTS
Follow the full ingest procedure documented in the skill:
1. Read `wiki/SCHEMA.md` first if you haven't this session — it may override default conventions. 2. Place the raw source in `raw/` if it isn't already there. Use a slugified filename. 3. Read the source. Chunk-read if it's large (over ~5000 words / a long PDF) — never load the whole thing into context if it would consume more than ~25% of the context window. 4. Briefly discuss the key takeaways with me before writing anything to the wiki — what stands out, what connects to existing pages, what's surprising. 5. Survey the wiki to identify which existing pages this source touches; read each candidate to confirm. 6. Write the source-summary page in `wiki/sources/`, surgically update touched entity/concept pages with `str_replace`, create new pages for new entities/concepts (each with at least one inbound link), update the index, append one line to `log.md`. 7. If `wiki/graph/ontology.yaml` exists and the ingest added or could add typed `graph.relationships[]`: add typed edges only when the source explicitly supports them (predicate, source-page slug, evidence quote, confidence, status). Then run `uv run --script "${CLAUDE_PLUGIN_ROOT}/skills/llm-wiki/scripts/wiki_graph_lint.py" "${PWD}/wiki"`, triage findings with me, and run `uv run --script "${CLAUDE_PLUGIN_ROOT}/skills/llm-wiki/scripts/wiki_graph_extract.py" "${PWD}/wiki"`. Append a ` graph: +N nodes, +M typed edges` sub-line under the ingest entry in `log.md`. 8. Tell me what you did when you're done: pages touched, pages created, contradictions flagged, follow-ups worth investigating.
If the wiki doesn't exist yet, suggest running `/wiki:init` first.
Turn PDFs, articles, transcripts, and notes into a shared wiki that your AI agents can search, cite, and keep up to date. Add a source once. Ask questions later. Keep the useful answers.
Other commands on praneybehl-llm-wiki.
- /graph
Compile, lint, or query the wiki's typed graph layer.
Open command - /init
Initialize a new LLM Wiki structure in this project (creates wiki/ and raw/ directories with templates).
Open command - /lint
Run a structural and semantic health check on the LLM Wiki.
Open command - /query
Query the LLM Wiki — answer a question from accumulated knowledge with citations.
Open command - /stats
Show wiki size, shape, and link density — and which scaling threshold the wiki is at.
Open command - /upgrade
Upgrade an existing LLM Wiki to the current plugin version (adds missing files idempotently; walks me through any SCHEMA.md merges by hand).
Open command