From llm-wiki
Use when the user invokes /llm-wiki:index or asks to regenerate the index.md files in an OKF bundle. Runs okf_index.py for deterministic bottom-up index generation.
How this skill is triggered — by the user, by Claude, or both
Slash command
/llm-wiki:indexThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Regenerates every `index.md` in an OKF bundle via a bottom-up walk. Groups entries by `type`, sorts alphabetically within each group, pulls `title` and `description` from each concept's frontmatter.
Regenerates every index.md in an OKF bundle via a bottom-up walk. Groups entries by type, sorts alphabetically within each group, pulls title and description from each concept's frontmatter.
/llm-wiki:index [path] [--llm]
path defaults to the nearest bundle root. --llm enables Gemini Flash directory descriptions (optional dep; default is deterministic).
Run:
python3 <plugin_root>/scripts/okf_index.py <bundle_root>
# With LLM directory descriptions:
python3 <plugin_root>/scripts/okf_index.py <bundle_root> --llm
The script prints the list of written index.md paths to stdout and a summary to stderr.
Then refresh bundle discovery, because a host briefing file may hold an inlined copy of the root index (see /llm-wiki:init):
python3 <plugin_root>/scripts/okf_discover.py <bundle_root> --sync
--sync refreshes blocks that already exist and never creates one, so it is safe to run unconditionally. If it reports skipped (no block …) for every file, discovery was never installed — offer to run okf_discover.py <bundle_root> (no flag) to install it.
Each directory with .md files (excluding index.md itself) gets an index.md with this format:
# BigQuery Dataset
* [Bitcoin Dataset](datasets/crypto_bitcoin.md) - Public blockchain data ETL'd from the Bitcoin network.
# BigQuery Table
* [Blocks](tables/blocks.md) - One row per block in the Bitcoin blockchain.
* [Transactions](tables/transactions.md) - One row per Bitcoin transaction.
# Subdirectories
* [references](references/index.md) - Contains 3 entries: metrics, joins, event_parameters.
/llm-wiki:visualize (the visualizer reads concept frontmatter, not indexes, but fresh indexes help navigation).Do not run during init — there are no concept docs yet, so it has nothing to catalog. (It is safe against the seeded root: regeneration preserves the okf_version frontmatter that marks the bundle root.) Run it once the first concept docs are added.
With --llm, Gemini Flash generates one-sentence directory descriptions (e.g. "Three join references between the four core Bitcoin tables."). Without --llm, descriptions are deterministic ("Contains 3 entries: metrics, joins, event_parameters."). For most bundles, the deterministic form is sufficient; upgrade with --llm when you want richer navigation text.
Alternatively, after running okf_index.py, you (Claude) can rewrite specific directory descriptions inline by editing the affected index.md — the regenerated descriptions are plain text, not locked.
npx claudepluginhub mlarkin00/plugins --plugin llm-wikiGuides creation and editing of skills using test-driven development with pressure scenarios and subagents to verify agent compliance.
Creates platform-native content for X, LinkedIn, TikTok, YouTube, and newsletters from source material. Adapts voice and format per platform while avoiding engagement bait and filler.