Tools

List the collections on lateano.com. Explain what each contains and include its path.

List the pages outside the site's searchable collections. Which one explains the API?

Search all fields for "caching". Return the strongest matches and say when more matched
than were returned.

Get the post whose slug is [post-slug], then read it and summarize its main argument.

List the tags. Find one related to [subject] and show how its documents are split by collection.

Get the page information for [route]. Group its outgoing references, referencing posts,
and related pages.

Read [route] as Markdown. Separate claims made by the page from your own inference.

Draft feedback about [route]: [message]. Classify its kind and severity, show the exact
submission, and wait for my approval before sending it.

---
title: "Tools"
description: "What each Tool accepts, returns, changes, and hands off to next."
url: "https://lateano.com/developers/mcp/tools"
author: "Colin Lateano"
---

# Tools

What each Tool accepts, returns, changes, and hands off to next.

Your client receives the executable input and output schemas from `tools/list`.
This page adds the choice the wire cannot make for you: when to use each Tool,
what changes, and which Tool normally follows.

Every Tool is non-destructive, advertises idempotence, and is limited to this
site's closed world. All but `submit_feedback` are read-only.

| Tool                | Use it to                             | Effect    |
| :------------------ | :------------------------------------ | :-------- |
| `list_collections`  | Learn the searchable sections         | Read-only |
| `list_pages`        | Enumerate pages outside search        | Read-only |
| `search_content`    | Find posts, terms, and demos          | Read-only |
| `get_post`          | Resolve one known post slug           | Read-only |
| `list_tags`         | Learn the site's subject vocabulary   | Read-only |
| `get_page_info`     | Read metadata and graph relationships | Read-only |
| `get_page_markdown` | Read what is written at one route     | Read-only |
| `submit_feedback`   | Record one reviewed message for Colin | Write     |

## `list_collections`

**Effect:** Read-only.

Use this first when you do not know how the searchable corpus is divided.

- **Input:** `limit` — optional `integer`, minimum 1. Omit it for the whole set.
- **Result fields:** `items[].collection`, `items[].title`, `items[].path`,
  `items[].count`, `count`, and `total`.
- **Next:** pass a returned `collection` to `search_content`.

```text
List the collections on lateano.com. Explain what each contains and include its path.
```

## `list_pages`

**Effect:** Read-only.

Use this for documentation, About, and index pages that no collection holds and
`search_content` cannot return.

- **Input:** `limit` — optional `integer`, minimum 1. Omit it for every page.
- **Result fields:** `items[].path`, `items[].title`, `items[].description`,
  `count`, and `total`.
- **Next:** pass a returned `path` to `get_page_markdown` or `get_page_info`.

```text
List the pages outside the site's searchable collections. Which one explains the API?
```

## `search_content`

**Effect:** Read-only.

Use this to search posts, glossary terms, and demos by words, collection, or tag.
At least one of those three must describe the question.

- **Input:** `match` — optional
  `{ text: string, in: title | body | description | aliases | tags | all }`.
- **Input:** `collection` — optional `writing | glossary | demos`.
- **Input:** `tag` — optional `string` returned by `list_tags`.
- **Input:** `limit` — optional `integer`, minimum 1 and maximum 25. The default
  limit is 10.
- **Result fields:** `items[].results[].path`, `items[].results[].canonical`,
  `items[].results[].title`, `items[].results[].description`,
  `items[].results[].collection`, `items[].results[].date`,
  `items[].results[].twin`, `items[].matched`, `items[].facets[].facet`,
  `items[].facets[].counts[].value`, `items[].facets[].counts[].count`, `count`,
  and `total`.
- **Limit:** when `matched` is larger than the returned rows, narrow the question.
  There is no next page after the maximum result set.
- **Next:** read a result's `path` with `get_page_markdown`.

```text
Search all fields for "caching". Return the strongest matches and say when more matched
than were returned.
```

## `get_post`

**Effect:** Read-only.

Use this when you already know a post's final URL segment. Use `search_content`
when the slug is uncertain.

- **Input:** `slug` — required `string`.
- **Result fields:** `items[].results[].path`, `items[].results[].canonical`,
  `items[].results[].title`, `items[].results[].description`,
  `items[].results[].collection`, `items[].results[].date`,
  `items[].results[].twin`, `items[].matched`, `items[].facets[].facet`,
  `items[].facets[].counts[].value`, `items[].facets[].counts[].count`, `count`,
  and `total`.
- **Next:** pass the returned result `path` to `get_page_markdown`.

```text
Get the post whose slug is [post-slug], then read it and summarize its main argument.
```

## `list_tags`

**Effect:** Read-only.

Use this before tag-filtered search when you need the real slug or want to learn
how a subject cuts across collections.

- **Input:** `limit` — optional `integer`, minimum 1. Omit it for the whole
  vocabulary.
- **Result fields:** `items[].slug`, `items[].title`, `items[].count`,
  `items[].byCollection`, `count`, and `total`.
- **Next:** pass a returned `slug` to `search_content` as `tag`.

```text
List the tags. Find one related to [subject] and show how its documents are split by collection.
```

## `get_page_info`

**Effect:** Read-only.

Use this when you know a page route and need metadata or graph relationships
rather than its prose.

- **Input:** `route` — required `string`, such as `/glossary` or `/` for the site
  root.
- **Result fields:** `page.path`, `page.canonical`, `page.title`,
  `page.description`, `page.indexing`, `page.collection`, `page.date`,
  `page.tags[]`, `page.aliases[]`, `page.twin`,
  `page.outgoingReferences[].path`, `page.outgoingReferences[].title`,
  `page.referencingPosts[].path`, `page.referencingPosts[].title`,
  `page.related[].path`, and `page.related[].title`.
- **Next:** follow a returned path, or read the page with `get_page_markdown`.

```text
Get the page information for [route]. Group its outgoing references, referencing posts,
and related pages.
```

## `get_page_markdown`

**Effect:** Read-only.

Use this to read a page after another Tool returns its route.

- **Input:** `route` — required `string`, including `/` for the site root.
- **Result:** plain Markdown text. When the page publishes a Markdown address, the
  result also carries that address as a resource link.
- **Limit:** a long page is cut. The result says where it stopped and links the
  complete published document; no Tool returns the remainder.
- **Next:** answer from the text, or use `get_page_info` for relationships.

```text
Read [route] as Markdown. Separate claims made by the page from your own inference.
```

## `submit_feedback`

**Effect:** Write.

Use this only after the user has reviewed the message. It records feedback for
Colin; it does not edit the site or return a reply.

- **Input:** `idempotency_key` — required `string (uuid)`.
- **Input:** `message` — required `string`.
- **Input:** `kind` — required `missing | broken | confusing | inaccurate | other`.
- **Input:** `severity` — required `high | medium | low`.
- **Input:** `about` — optional `string` route.
- **Result fields:** `about` and `receivedAt`.
- **Retry:** reuse the same UUID only for the same submission. Generate a new UUID
  when the message, kind, severity, or route changes.
- **Next:** report what was recorded and when. No Tool reads the private message back.

```text
Draft feedback about [route]: [message]. Classify its kind and severity, show the exact
submission, and wait for my approval before sending it.
```