> ## Documentation Index
> Fetch the complete documentation index at: https://saydo-5cd0e3d7.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# /wiki-search

> Search across all wiki articles by keyword or phrase.

`/wiki-search` finds articles and sections that match your keywords. It works at any scale without external dependencies, searching topic names first and then full article content if needed.

## Command

```bash theme={null}
/wiki-search <query>
```

Multi-word queries are supported — no quotes needed.

```bash theme={null}
/wiki-search retention experiments
/wiki-search Day 1 activation
/wiki-search push notification tradeoffs
```

If you run `/wiki-search` with no argument, you'll be prompted to enter a query.

## How search works

Search uses a two-tier approach to stay fast while still finding deeply buried content:

**Tier 1 — Index scan:** The command reads `INDEX.md` first and checks topic names and one-line summaries for your keywords. This is usually enough to answer "which topic covers X?" and returns results immediately.

**Tier 2 — Full article search:** If the index scan doesn't find a match, the command searches across all files in `topics/` and `concepts/` using a full grep. Results include the matching section heading and surrounding lines so you can see the match in context.

## Result format

```
Found 3 matches for "retention experiments":

### Topics
- **Retention** [coverage: high — 15 sources] — "...D1 experiments in Q1 showed..."
  → wiki/topics/retention.md

- **Experiment Design** [coverage: medium — 4 sources] — "...retention as primary metric..."
  → wiki/topics/experiment-design.md

### Concepts
- **Speed vs Quality Tradeoff** — "...retention impact weighed against..."
  → wiki/concepts/speed-vs-quality.md
```

Coverage indicators are included in every result so you know how much to trust the wiki excerpt versus reading the raw source files directly:

* `[coverage: high]` — 5 or more sources; trust this section
* `[coverage: medium]` — 2–4 sources; good overview, check raw files for specifics
* `[coverage: low]` — 0–1 sources; read the raw files listed in the article's Sources section

## When there are no matches

If nothing is found, the command suggests related topics that might be relevant and recommends next steps:

* `/wiki-query` for questions that require connecting multiple topics
* `/wiki-ingest` if you think the knowledge should be in the wiki but hasn't been added yet

## When to use `/wiki-query` instead

`/wiki-search` is keyword-based — it finds where a term appears. If you want a synthesized answer that draws from multiple topics, use `/wiki-query` instead.

```bash theme={null}
# Finding where a topic lives → use /wiki-search
/wiki-search retention

# Getting an answer that connects topics → use /wiki-query
/wiki-query what do we know about retention drop-off after Day 1?
```