A command for managing indexes that point to large datasets kept outside the wiki. The wiki stores the catalogue and access information, while the actual data remains elsewhere.

Use it to create, update, and maintain dataset manifests—files that describe where datasets are and how they are organized.

It keeps oversized or unsuitable data out of the wiki without losing a clear way to find and use it.

dataset — Command by nvk

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add commands/nvk/llm-wiki/dataset

Clone the repo

git clone --depth 1 https://github.com/nvk/llm-wiki

Or install claude-plugin, the plugin that ships this one along with the rest of its 1 skill, 28 commands.

Per session 30 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 2,358 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5	$0.00030	$0.02358
Opus 5	$0.00015	$0.01179
Sonnet 5	$0.00006	$0.00472
Haiku 4.5	$0.00003	$0.00236

Measured 3d ago against content hash 845b1ba15151, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

dataset scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

claude-plugin/commands/dataset.md · 176 lines

How it starts

The opening of the file, as written. The whole thing — 176 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Your task

Resolve the wiki. Do NOT broadly search the filesystem — follow these steps:

Read $HOME/.config/llm-wiki/config.json. If it has hub_path, expand leading ~ only (not tildes in com~apple~CloudDocs) and prefer that path; use resolved_path only as a fallback cache when the expanded hub_path is unavailable and resolved_path is initialized. If config has only resolved_path, use it. If the configured path can be statted but reading wikis.json or listing topics/ fails with Operation not permitted, stop and ask the user to grant Full Disk Access/iCloud Drive access to the launcher; do not fall back to ~/wiki or resolved_path. Do not write machine-specific resolved_path into shared configs.
If no config → read $HOME/wiki/_index.md. If it exists → HUB = $HOME/wiki. If nothing found, ask the user where to create the wiki.
Wiki location (first match): --local → .wiki/ in CWD; --wiki <name> → HUB/wikis.json lookup with portable path resolution (<HUB>, ~, absolute, or HUB-relative); if the registry path is stale, fall back to HUB/topics/<name>; CWD has .wiki/ → use it; else → HUB.
Read <wiki>/_index.md to verify. If missing → stop with "No wiki found. Run /wiki init first."

Archive rule: dataset commands operate on active topic wikis by default. If a named target is archived, stop and ask the user to restore it or rerun with --include-archived. When explicitly included, list/update manifests only inside that archived topic path and label the result as archived. This is separate from dataset manifest status: archived.

After resolving the wiki, read the dataset reference at skills/wiki-manager/references/datasets.md, then run the requested subcommand.

Dataset manifests are for large or external data that should not be copied into raw/ or wiki/. The wiki stores pointers, schema notes, small samples, profiles, query recipes, and provenance. Actual datasets remain at their original filesystem path, object store, URL, database, or archive.

Be opinionated about whether a dataset manifest is warranted:

Good fit: large, mutable, remote, compressed, binary, database-backed, or query-oriented data that should stay outside markdown.
Too small: a short CSV/JSON/text file that can be ingested as one immutable raw/data/ source.
Too operational: a next-action queue about a corpus belongs in inventory, with an optional link to a dataset manifest.
Too broad: many independent sources/pages should usually be ingest-collection, not a dataset manifest.

Before larger pivots, preview the shape: one dataset manifest plus any linked inventory record, sample/profile/query notes, and what data will remain external. Do not ask the user to approve a big migration without showing this sample.

Read the full file on GitHub · 176 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

3d ago First seen · 176 lines · 30 tokens per session scan A 845b1ba15151

Subscribe to this mod's changes

dataset is a command published in the GitHub repository nvk/llm-wiki (1,175 stars, last pushed 6d ago), licensed MIT. It adds 30 tokens to every session and 2,358 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Data and AI LLM Research agentic-ai agentic-skills agentic-workflow claude-code codex

dataset

Getting it into your agent

What it costs to keep this loaded

Grade A, and why

Nothing flagged

How it starts

Your task

What this file has done since we first saw it

Other commands, from other repositories

cost-analyze

implement

loop

project

status

ask