Skip to content
Back to skills

Wiki All

ASecurity

Run the complete llmwiki pipeline from scratch — init, sync, graph, build, lint, and serve. Use when the user says "run everything", "full pipeline", "wiki-all", or wants to rebuild the entire wiki from session history.

  • 383 stars
  • 0 votes
  • 0 copies
  • 3 views
  • Added May 26, 2026
developmentpythongobashnode

Security analysis

A100/100

Scanned May 27, 2026

npx -y skills add Pratiyush/llm-wiki --skill wiki-all --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Wiki All?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Wiki All
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/pratiyush-wiki-all/badge)](https://www.skillsdirectory.com/skills/pratiyush-wiki-all)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: wiki-all
description: Run the complete llmwiki pipeline from scratch — init, sync, graph, build, lint, and serve. Use when the user says "run everything", "full pipeline", "wiki-all", or wants to rebuild the entire wiki from session history.
---

# wiki-all

## What this skill does

Runs the complete llmwiki pipeline in the correct sequence:

1. **init** — scaffold `raw/`, `wiki/`, `site/` directories
2. **sync** — convert `.jsonl` sessions to markdown
3. **graph** — build Graphify AI knowledge graph (communities, god nodes, hyperedges)
4. **build** — compile `wiki/` markdown into `site/` HTML
5. **lint** — run 14 quality rules, report issues
6. **serve** — start local server at http://127.0.0.1:8765

## Steps

Run each step in order. Report the result of each before proceeding to the next.

### Step 1: Init

```bash
python3 -m llmwiki init
```

Report what was created or confirmed.

### Step 2: Sync

```bash
python3 -m llmwiki sync --no-auto-build --no-auto-lint
```

Report: how many sessions converted, how many unchanged, any errors.

If new sessions were converted, follow the **Ingest Workflow** from `CLAUDE.md` for each new file (create/update entity + concept pages, cross-link, update index).

### Step 3: Graph

```bash
python3 -m llmwiki graph
```

Report: node count, edge count, community count, top 5 connected nodes.

Ask the user: **"Export graph to Obsidian vault?"** If yes:

```python
python3 -c "
from llmwiki.graphify_bridge import export_to_obsidian
from pathlib import Path
export_to_obsidian(Path.home() / 'Documents/Obsidian Vault/Temp/Graph')
"
```

### Step 4: Build

```bash
python3 -m llmwiki build
```

Report: HTML file count, total size, export formats written.

### Step 5: Lint

```bash
python3 -m llmwiki lint
```

Report: total issues by severity (errors, warnings, info). Highlight any errors that need immediate attention.

### Step 6: Serve

Ask the user: **"Start the local server?"** If yes:

```bash
python3 -m llmwiki serve --open
```

Report the URL: http://127.0.0.1:8765

## After completion

Report a summary table:

| Step | Result |
|------|--------|
| init | N dirs, M seed files |
| sync | N converted, M unchanged |
| graph | N nodes, M edges, K communities |
| build | N HTML files, M MB |
| lint | N issues (E errors, W warnings) |
| serve | http://127.0.0.1:8765 |

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.