How do I install the Ingest skill?

Run npx skillmds@latest add protonium-labs/ingest in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Ingest safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Ingest?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Ingest free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Ingest?

protonium-labs (@protonium-labs) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Ingest by protonium-labs — Agent Skill

Name: Ingest
Author: protonium-labs

File contents

Task

Execute the requested action: $ARGUMENTS

Calling rule: /ingest <action> <arguments>.

If the requested action is ambiguous or contradictory, do NOT guess: flag it and ask for clarification before executing.

If action is empty

Describe this skill: its purpose and the actions table from "overview". Do not execute anything.

If action is "overview" (or unknown)

Argument	Action
`all`	Process every unprocessed file in `vault/sources/`
`file <name>`	Process one specific source file
`overview`	This table

If the action was unknown, say so first, then show the table.

If action is "all" or "file"

Scope:
- all: list vault/sources/ and compare against vault/log.md ingest entries to find unprocessed files. Large batches: one source at a time, report progress.
- file <name>: match the name in vault/sources/; if missing or empty, list the folder and ask.
Read each source fully. PDFs, images, transcripts, articles alike.
Integrate, don't just summarize, per source:
- Update existing pages where the topic exists; create new pages only for genuinely new topics (one topic per page).
- Add YAML frontmatter (tags, created, updated, source) and [[wiki links]] to every related page. A single source may touch 10-15 pages.
- Route by content: user-related → vault/me/, project-related → vault/projects/<name>/, general knowledge → vault/knowledge/.
- Flag contradictions explicitly in the page and in the report; never silently overwrite.
Never modify the source file itself (CLAUDE.md hard rule).

After any action

One log entry per source: ## [YYYY-MM-DD HH:MM] /ingest | <source> -> pages touched.
Update vault/index.md with new pages.
Report: sources processed, pages created/updated, contradictions flagged.

Notes

All pages must pass markdownlint.

---name: ingestdescription: Process raw files from vault/sources/ into linked wiki pages. Actions - all (every unprocessed source), file (one specific source), overview.---# TaskExecute the requested action: $ARGUMENTSCalling rule: `/ingest <action> <arguments>`.If the requested action is ambiguous or contradictory, do NOT guess: flag it and ask for clarification before executing.---## If action is emptyDescribe this skill: its purpose and the actions table from "overview". Do not execute anything.## If action is "overview" (or unknown)| Argument | Action || :--- | :--- || `all` | Process every unprocessed file in `vault/sources/` || `file <name>` | Process one specific source file || `overview` | This table |If the action was unknown, say so first, then show the table.---## If action is "all" or "file"1. Scope: - `all`: list `vault/sources/` and compare against `vault/log.md` ingest entries to find unprocessed files. Large batches: one source at a time, report progress. - `file <name>`: match the name in `vault/sources/`; if missing or empty, list the folder and ask.2. **Read each source fully.** PDFs, images, transcripts, articles alike.3. **Integrate, don't just summarize**, per source: - Update existing pages where the topic exists; create new pages only for genuinely new topics (one topic per page). - Add YAML frontmatter (`tags`, `created`, `updated`, `source`) and `[[wiki links]]` to every related page. A single source may touch 10-15 pages. - Route by content: user-related → `vault/me/`, project-related → `vault/projects/<name>/`, general knowledge → `vault/knowledge/`. - **Flag contradictions** explicitly in the page and in the report; never silently overwrite.4. **Never modify the source file itself** (CLAUDE.md hard rule).---## After any action- One log entry per source: `## [YYYY-MM-DD HH:MM] /ingest | <source> -> pages touched`.- Update `vault/index.md` with new pages.- Report: sources processed, pages created/updated, contradictions flagged.## Notes- All pages must pass markdownlint.

Frequently asked questions

Ingest