po4yka/llm-wiki-skills / skills/llm-wiki-zero-to-working-wiki
llm-wiki-zero-to-working-wiki
VerifiedCreate a first useful LLM-Wiki from zero. Use when the user wants a zero-to-working-wiki walkthrough, first vault, first source ingest, first reusable query, first lint report, or a visible demo loop before a larger rollout; route full workflow installs to llm-wiki-setup.
Zero to Working Wiki
Goal
Get from an empty folder to a working, reviewable LLM-Wiki loop in one small session.
When to use
Use this skill when the user wants:
- a first local LLM-Wiki vault;
- a product demo that shows value before a migration;
- one raw source turned into a draft wiki page;
- one reusable answer saved back into the wiki;
- one lint report that creates a maintenance queue.
Inputs
- Target folder.
- Agent target: Claude Code, Codex, Cursor, OpenCode or another compatible agent.
- Write mode: direct local writes, dry-run patch, or PR-based.
- Privacy constraints.
- Optional first source text or file path.
Procedure
1. Preflight
Check:
- the target folder exists or can be created;
- git is initialized or the user accepts initializing it;
- no sensitive source should leave the machine;
- writes are reviewable through
git diff; - the user wants a tiny first loop, not a bulk migration.
If the user has an existing vault or repository, keep this run scoped to one source and one question.
2. Create the minimal vault shape
Create only missing directories:
raw/sources/
raw/assets/
inbox/
wiki/sources/
wiki/entities/
wiki/concepts/
wiki/queries/
wiki/synthesis/
_agent/reports/
_meta/schemas/
Create minimal instruction and navigation files if absent, starting from the bundled starters:
AGENTS.mdfromreferences/templates/vault/AGENTS.md;CLAUDE.mdfromreferences/templates/vault/CLAUDE.mdwhen the user uses Claude Code;wiki/index.mdfromreferences/templates/wiki/index.md;wiki/log.mdfromreferences/templates/wiki/log.md.
Whatever the starting point, the instruction files must state:
- do not edit
raw/files; - generated pages start as
status: draft; - human synthesis sections are protected;
- durable changes update
wiki/index.mdandwiki/log.md; - bulk writes require approval.
3. Add one source
If the user did not provide a source, create a small sample source under raw/sources/example-source.md:
# Example source
A living LLM-Wiki preserves raw sources, creates reviewable summaries, saves useful answers back into durable pages and runs periodic lint checks.
Do not overwrite an existing source with the same name.
4. Run first ingest
Create a draft source page under wiki/sources/ with:
- frontmatter;
- source path;
- extracted claims;
- entities and concepts;
- ambiguities and caveats;
- links created;
- review-required status.
Update wiki/index.md and append wiki/log.md.
5. Run first query
Answer one question from the wiki:
What maintenance loop does this source recommend?
If the answer is reusable, save it as a draft page under wiki/queries/. If not, explain why it was not saved.
6. Run first lint report
Create a report at _agent/reports/YYYY-MM-DD-first-lint.md with:
- summary;
- broken links;
- orphan pages;
- provenance gaps;
- stale pages;
- contradictions;
- suggested patches;
- next action.
Do not silently change truth-bearing pages while linting.
7. Show the product result
End by showing:
- files created or changed;
- what a human can now read without an agent;
- what remains draft or review-required;
- the next useful skill to run.
Output
## Zero-to-working-wiki summary
## Files created or changed
## First source
## First reusable answer
## First lint queue
## Review-required items
## Next action
Safety gates
- Do not start a bulk migration in this skill.
- Do not mark generated pages
verified. - Do not overwrite existing instruction, index, log or source files without showing a diff plan.
- Do not ingest private material into cloud tools without explicit approval.
- If the first source contains time-sensitive claims, mark them
re-verifyinstead of treating them as current. - Do not edit protected human synthesis sections.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| planning-and-task-breakdownaddyosmani/agent-skillsBreaks work into ordered tasks. Use when you have a spec or clear requirements and need to break work into implementable tasks. Use when a task feels too large to start, when you need to estimate scope, or when parallel work is possible. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| screenpipethedotmack/claude-memSearch screen and audio history recorded by Screenpipe and optionally save selected context into claude-mem. Use when the user asks to recall desktop activity, a meeting, or something seen in another app through Screenpipe. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| contact-widgetnexu-io/open-designSelf-contained floating chat widget with welcome screen, social links, meeting button, and message input. Single HTML file, zero dependencies. | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |
| meeting-notesnexu-io/open-designMeeting notes page — title bar with attendees, agenda checklist, decisions block, action items table with owners + dates, and a "next meeting" footer. Use when the brief mentions "meeting notes", "minutes", "1:1 notes", "all-hands recap", or "会议纪要". | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |