# Hacker News Comment Search API

Hacker News

Returns the public Hacker News comments that match your words, ranked by relevance or newest first. Filter by author, by the story they were posted on, or by date. Each comment has its text as plain text, author, posting time, reply count, the comment it replies to and the title and link of its story. [Overview and pricing](https://socialtap.dev/apis/hackernews/search-comments/index.md)

GET `/v1/hackernews/search_comments`

## Example

cURL:

```shell
curl -G 'https://socialtap.dev/v1/hackernews/search_comments' \
  --data-urlencode 'query=rust borrow checker' \
  -H "x-api-key: $SOCIALTAP_API_KEY"
```

TypeScript SDK:

```typescript
import { SocialTap } from "socialtap";

const client = new SocialTap(); // apiKey defaults to process.env.SOCIALTAP_API_KEY

const res = await client.hackernews.searchComments.list({ query: "rust borrow checker" });
console.log(res.data);
console.log(res.credits.charged, res.meta.request_id);
```

Node.js (fetch):

```javascript
const params = new URLSearchParams({
  query: "rust borrow checker",
});
const res = await fetch(`https://socialtap.dev/v1/hackernews/search_comments?${params}`, {
  headers: { "x-api-key": process.env.SOCIALTAP_API_KEY ?? "" },
});
const body = await res.json();
if (!res.ok) throw new Error(`${body.error.code}: ${body.error.message}`);
const { data, page, meta } = body;
console.log(data, meta.credits_charged);
```

Python (requests):

```python
import os

import requests

res = requests.get(
    "https://socialtap.dev/v1/hackernews/search_comments",
    params={"query": "rust borrow checker"},
    headers={"x-api-key": os.environ["SOCIALTAP_API_KEY"]},
)
body = res.json()
if not res.ok:
    raise RuntimeError(f"{body['error']['code']}: {body['error']['message']}")
data, page, meta = body["data"], body["page"], body["meta"]
print(data, meta["credits_charged"])
```

Go (net/http):

```go
package main

import (
	"encoding/json"
	"fmt"
	"log"
	"net/http"
	"net/url"
	"os"
)

func main() {
	q := url.Values{}
	q.Set("query", "rust borrow checker")
	req, err := http.NewRequest("GET", "https://socialtap.dev/v1/hackernews/search_comments?"+q.Encode(), nil)
	if err != nil {
		log.Fatal(err)
	}
	req.Header.Set("x-api-key", os.Getenv("SOCIALTAP_API_KEY"))
	res, err := http.DefaultClient.Do(req)
	if err != nil {
		log.Fatal(err)
	}
	defer res.Body.Close()
	var body struct {
		Data  json.RawMessage `json:"data"`
		Page  json.RawMessage `json:"page"`
		Meta  map[string]any  `json:"meta"`
		Error *struct {
			Code    string `json:"code"`
			Message string `json:"message"`
		} `json:"error"`
	}
	if err := json.NewDecoder(res.Body).Decode(&body); err != nil {
		log.Fatal(err)
	}
	if body.Error != nil {
		log.Fatalf("%s: %s", body.Error.Code, body.Error.Message)
	}
	fmt.Println(string(body.Data), body.Meta["credits_charged"])
}
```

200 OK:

```json
{
  "data": [
    {
      "approximate_fields": [],
      "author": {
        "avatar_url": null,
        "display_name": null,
        "id": "Animats",
        "is_verified": null,
        "url": "https://news.ycombinator.com/user?id=Animats",
        "username": "Animats"
      },
      "edited_at": null,
      "id": "9948436",
      "is_by_post_author": null,
      "is_edited": null,
      "is_liked_by_author": null,
      "is_pinned": null,
      "metrics": {
        "like_count": null,
        "reply_count": 0
      },
      "parent_id": "9947256",
      "platform": "hackernews",
      "platform_specific": {
        "hackernews": {
          "depth": null,
          "is_dead": false,
          "is_deleted": false,
          "story": {
            "comment_count": null,
            "link_url": "https://www.justsoftwaresolutions.co.uk/cplusplus/shared-ptr-secret-constructor.html",
            "points": null,
            "published_at": null,
            "title": "Std::shared_ptr's secret constructor"
          }
        }
      },
      "post_id": "9946625",
      "published_at": "2015-07-25T18:14:21Z",
      "text": "\"Interesting, but in most circumstances I think this falls into the \"too clever\" category. As with most of C++'s weirder features, you really only want to use them when you absolutely have to.\"\n\nThe C++ committee has been trying to get ownership semantics to work right for about …",
      "unknown_fields": [],
      "url": "https://news.ycombinator.com/item?id=9948436"
    },
    {
      "approximate_fields": [],
      "author": {
        "avatar_url": null,
        "display_name": null,
        "id": "catnaroek",
        "is_verified": null,
        "url": "https://news.ycombinator.com/user?id=catnaroek",
        "username": "catnaroek"
      },
      "edited_at": null,
      "id": "9858523",
      "is_by_post_author": null,
      "is_edited": null,
      "is_liked_by_author": null,
      "is_pinned": null,
      "metrics": {
        "like_count": null,
        "reply_count": 0
      },
      "parent_id": "9853004",
      "platform": "hackernews",
      "platform_specific": {
        "hackernews": {
          "depth": null,
          "is_dead": false,
          "is_deleted": false,
          "story": {
            "comment_count": null,
            "link_url": "http://devblog.avdi.org/2015/07/08/ruby-is-defined-by-terrible-tools/",
            "points": null,
            "published_at": null,
            "title": "Ruby Is Defined by Terrible Tools"
          }
        }
      },
      "post_id": "9851685",
      "published_at": "2015-07-09T15:42:52Z",
      "text": "You're confusing \"sophisticated\" with \"annoying or difficult to use\". Boilerplate generators are annoying, and Valgrind is difficult to use, but they're unrefined: they solve problems that are best avoided than solved.\n\nA better measure of sophistication is how much bang you get …",
      "unknown_fields": [],
      "url": "https://news.ycombinator.com/item?id=9858523"
    },
    {
      "approximate_fields": [],
      "author": {
        "avatar_url": null,
        "display_name": null,
        "id": "Aatch",
        "is_verified": null,
        "url": "https://news.ycombinator.com/user?id=Aatch",
        "username": "Aatch"
      },
      "edited_at": null,
      "id": "9783186",
      "is_by_post_author": null,
      "is_edited": null,
      "is_liked_by_author": null,
      "is_pinned": null,
      "metrics": {
        "like_count": null,
        "reply_count": 1
      },
      "parent_id": "9781361",
      "platform": "hackernews",
      "platform_specific": {
        "hackernews": {
          "depth": null,
          "is_dead": false,
          "is_deleted": false,
          "story": {
            "comment_count": null,
            "link_url": "http://blog.rust-lang.org/2015/06/25/Rust-1.1.html",
            "points": null,
            "published_at": null,
            "title": "Rust 1.1 Stable, the Community Subteam, and RustCamp"
          }
        }
      },
      "post_id": "9779757",
      "published_at": "2015-06-26T09:38:05Z",
      "text": "Eh, at the end of the day, they're different languages and are following different paths. Sure they're superficially similar, but that's not all to it.\n\nPython and Ruby are pretty similar languages overall, but neither has \"killed\" the other. Why? Because despite their similiarit…",
      "unknown_fields": [],
      "url": "https://news.ycombinator.com/item?id=9783186"
    }
  ],
  "page": {
    "has_more": true,
    "next_cursor": "c1.EXAMPLE_CURSOR",
    "next_cursor_expires_at": "2026-10-26T12:00:00Z",
    "end_reason": null
  },
  "meta": {
    "request_id": "req_3f9c2a7d1e8b4c6a9f0e2d4b6a8c0e1f",
    "platform": "hackernews",
    "source": "api",
    "is_cached": false,
    "credits_charged": 1,
    "latency_ms": 1000,
    "fetched_at": "2026-10-25T12:00:00Z",
    "input": null,
    "credits_remaining": 9999,
    "client_request_id": null,
    "warnings": []
  }
}
```

Long arrays and strings are shortened for display; real responses are complete.

> [!NOTE]
>
> **Good to know**
>
> - Results come from HN Search, the search Hacker News itself links to, which copies every comment within minutes.
> - Hacker News hides comment points from everyone but their author, so like_count is null.
> - sort=relevance serves the first thousand matching comments, then ends with platform_limit. sort=newest walks back through every match.
> - Comments Hacker News killed or that were deleted are not in the results.
> - Words in Chinese, Japanese or Korean match only where they stand apart as words, so they find few comments.

## Authorizations

`x-api-key`stringheaderrequired

Your API key. [Authentication](https://socialtap.dev/docs/authentication/index.md)

## Query parameters

| Name | Type | Description | Example |
| --- | --- | --- | --- |
| `query` required | `string`<br>max 500 chars | Words to find comments by, as HN Search reads them: any word can match with a small typo, a phrase in quotes matches exactly, and a leading - leaves a word out (rust borrow checker, "memory safety" -c++). Case does not matter. A link to a comment search on HN Search (https://hn.algolia.com/?q=rust&type=comment&sort=byDate) runs that search: its words, sort, dates, author:name and story:id fill the params you leave out. | `rust borrow checker` |
| `sort` optional | `string` | Order of the results. relevance (default): HN Search's ranking, the best match of the words first. newest: newest first, back through every matching comment.<br>One of `relevance`, `newest` | — |
| `author` optional | `string`<br>max 300 chars | Only comments written by this user: a username (case-sensitive, as on Hacker News) or a profile link. | `pg` |
| `story` optional | `string`<br>max 300 chars | Only comments on this story, replies included: a story's link (news.ycombinator.com/item?id=8863) or its item ID. A comment's ID matches nothing: Hacker News files every comment under its story. | `8863` |
| `search_in` optional | `string` | Where the words must appear: all (default: the comment's text and its author's username) or text (the text alone).<br>One of `all`, `text` | `text` |
| `since` optional | `string`<br>date-time | Only items at or after this time: RFC 3339 (2026-10-01T12:00:00Z) or a date (2026-10-01, midnight UTC). On newest-first lists, paging stops when it is reached, so later pages are not fetched or charged. By the comment's posting time, to the second. | `2026-10-01` |
| `until` optional | `string`<br>date-time | Only items before this time (exclusive): RFC 3339 or a date (midnight UTC). By the comment's posting time, to the second. | `2026-10-08` |
| `limit` optional | `integer`<br>min 1<br>max 100 | Items per page, 1 to the endpoint's page size (its default and maximum; larger values are reduced to it). Omit it: every page costs the same, so a smaller limit means more pages to pay for. | `100` |
| `cursor` optional | `string` | page.next_cursor from the previous response. Send the same other parameters; omit to start from the first page. | — |

`query` takes Hacker News search links; another kind of link, or another platform's, fails free with [`invalid_input`](https://socialtap.dev/docs/errors/index.md#invalid_input). Unknown parameters are rejected with a suggestion, never ignored ([Requests](https://socialtap.dev/docs/requests/index.md)). `limit` and `cursor` page through the list: see Pagination.

**Parameters every endpoint accepts**

| Name | Type | Description | Example |
| --- | --- | --- | --- |
| `max_age` optional | `integer`<br>min 0 | Accept a cached response up to this many seconds old. Cache hits cost 0 credits. Omit to always fetch fresh. The Cache-Control: max-age=N request header is equivalent; max_age wins when both are sent. | — |
| `fields` optional | `string` | Return only these fields of each item: comma-separated paths in the schema's notation (id,text,metrics.view_count,segments\[\].text; a parent path keeps the whole object). Unknown paths are rejected. Applied after the cache, so it changes neither the price nor cache hits; page and meta are always returned whole. | — |

## Response

### 200

`{data, page, meta}`. `data` is an array of up to 100 [comment](https://socialtap.dev/resources/comment/index.md) objects, HN Search's ranking: the best match of the words first; `page` says how to get the next page; `meta` describes the request ([meta](https://socialtap.dev/docs/reference/index.md#meta)). Fields only Hacker News has are under `platform_specific.hackernews`.

| Field | Type | Description |
| --- | --- | --- |
| `data[].platform` | `string` | Platform ID (tiktok, youtube, …). |
| `data[].id` | `string` | Platform-native comment ID (string; Reddit with its t1_ prefix). |
| `data[].post_id` | `string` | ID of the post commented on (equals post.id). |
| `data[].parent_id` | `string` or `null` | ID of the comment this replies to; null for top-level comments. |
| `data[].url` | `string` (uri) or `null` | Permalink, when the platform has one. |
| `data[].text` | `string` | Comment body as plain text. |
| `data[].author` | `object` | The commenting account. An account the platform no longer names (a deleted account) has a null id and username and the platform's label as display_name (Reddit: \[deleted\]). |
| `data[].author.id` | `string` or `null` | Platform-native account ID; null when only the handle is known. |
| `data[].author.username` | `string` or `null` | Handle without the @; null when the account has none. |
| `data[].author.display_name` | `string` or `null` | Display name shown on the profile. |
| `data[].author.url` | `string` (uri) or `null` | Canonical profile URL. |
| `data[].author.avatar_url` | `string` (uri) or `null` | Profile picture URL. Platform CDN URLs can expire within hours. |
| `data[].author.is_verified` | `boolean` or `null` | Platform verification badge; null where the platform has no badge. |
| `data[].published_at` | `string` (date-time) or `null` | When the comment was posted, RFC 3339 UTC. Where the platform shows only a relative time ("3 weeks ago"), the latest time that label allows (the fetch time minus the stated amount), listed in approximate_fields. |
| `data[].edited_at` | `string` (date-time) or `null` | When the comment was last edited, RFC 3339 UTC; null when never edited or when the platform marks edits without a time (see is_edited). |
| `data[].is_edited` | `boolean` or `null` | The comment was edited after posting (the platform's edited mark); null when the platform does not say. |
| `data[].is_pinned` | `boolean` or `null` | Pinned by the post's author. |
| `data[].is_liked_by_author` | `boolean` or `null` | Liked or hearted by the post's author. |
| `data[].is_by_post_author` | `boolean` or `null` | Written by the post's author (the creator's own comment, marked as such by the platform). |
| `data[].metrics` | `object` | Counts at fetch time; null = hidden or not exposed, 0 = zero. |
| `data[].metrics.like_count` | `integer` or `null` | Likes or upvotes; never negative. |
| `data[].metrics.reply_count` | `integer` or `null` | Direct replies. |
| `data[].platform_specific` | `object` | Fields only this platform has, under its ID (platform_specific.\<platform>); each null when unknown. |
| `data[].platform_specific.hackernews` | `object` or `null` | Hacker News' own comment fields. |
| `data[].platform_specific.hackernews.depth` | `integer` or `null` | How deep in the story's thread the comment is: 0 for a comment on the story, 1 for a reply to one, and so on. Null where the endpoint does not read the thread. |
| `data[].platform_specific.hackernews.is_deleted` | `boolean` or `null` | The comment's author deleted it. Hacker News keeps it in the thread as \[deleted\] while it has replies: text is then empty and author is \[deleted\]. |
| `data[].platform_specific.hackernews.is_dead` | `boolean` or `null` | Hacker News killed the comment (flagged by users or caught by its filters), so its text is hidden from logged-out readers: text is then empty. It stays in the thread while it has replies. |
| `data[].platform_specific.hackernews.story` | `object` or `null` | The story the comment is on (post_id), as far as the endpoint read it; null in a list of a thread's comments, which all share one story. |
| `data[].platform_specific.hackernews.story.title` | `string` or `null` | The story's title, as Hacker News shows it beside the comment. |
| `data[].platform_specific.hackernews.story.link_url` | `string` (uri) or `null` | The page the story links to; null for a text story (Ask HN) and where the endpoint read only the title. |
| `data[].platform_specific.hackernews.story.points` | `integer` or `null` | The story's points at fetch time. |
| `data[].platform_specific.hackernews.story.comment_count` | `integer` or `null` | Comments on the story at fetch time, replies included. |
| `data[].platform_specific.hackernews.story.published_at` | `string` (date-time) or `null` | When the story was submitted, RFC 3339 UTC. |
| `data[].approximate_fields` | `string[]` | JSON paths of values rounded by the platform or parsed from display text. A list path (external_links) means the list may be incomplete. \[\] when all exact. |
| `data[].unknown_fields` | `string[]` | JSON paths of lists whose \[\] means unknown rather than none (a source that cannot see them); null already means unknown for every other field. \[\] when every list is known. |

not_found and private_or_unavailable are item-level: the request was fine, the item can't be returned.

### 400

400 Bad Request: 2 error codes. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `invalid_input` | 400 | No | A parameter is missing or malformed, or the link is not one this endpoint takes. | Fix the parameter named in param as the message says. For a link of the wrong kind, call an endpoint listed in details.endpoints, when there is one. |
| `invalid_cursor` | 400 | No | The cursor is malformed or expired, belongs to another endpoint, or came with other parameters than the request that returned it. | Repeat the original request without cursor to start the list again. |

### 401

401 Unauthorized: 1 error code. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `unauthorized` | 401 | No | The x-api-key header is missing, or its key is unknown or revoked. On the dashboard's routes: the session cookie is missing or expired. | Send a valid API key in the x-api-key header. No key yet? A person gets one at details.signup_url; an AI agent follows details.agent_auth_url. |

### 402

402 Payment Required: 2 error codes. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `insufficient_credits` | 402 | No | The account does not have enough credits for this request. | Have a person buy credits at details.purchase_url (or make them a checkout link with details.checkout), then repeat the request. Do not retry automatically. |
| `spend_limit_reached` | 402 | No | The API key reached its spend limit for this period. The account may still have credits. | Wait until details.resets_at, or raise the key's spend limit (PATCH /v1/api_keys/{id}). Do not retry automatically. |

### 403

403 Forbidden: 2 error codes. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `forbidden` | 403 | No | The API key is valid but may not make this request, and details.reason says why: the account is suspended, the key lacks the route's scope, has expired, or was used from an IP outside its allowlist, or a rotated key's previous secret was used for a call that needs the admin scope. A running job stops with it too when its account is suspended or the key that created it is revoked or expires. It is about the key, never the item: a private item is private_or_unavailable, also 403, so branch on the code. | Act on details.reason. suspended: contact [support](mailto:support@socialtap.dev). scope: use a key with details.required_scope. previous_secret: use the new secret. key_expired or key_revoked: create a new key. ip_not_allowed: call from an allowed IP. |
| `private_or_unavailable` | 403 | No | The item exists but is not public: it is private, age- or region-restricted, or needs a login to view. It shares 403 with forbidden (which is about the key), so branch on the code. | Only public items can be fetched. Do not retry. |

### 404

404 Not Found: 1 error code. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `not_found` | 404 | No | The item does not exist or was deleted. On GET /v1/requests/{id}/response: the response was not kept (details.reason says why). | Check the link or ID. Do not retry. |

### 429

429 Too Many Requests: 1 error code. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `rate_limited` | 429 | Yes, later | The account passed its runaway-protection ceiling: a high per-account limit that stops a broken client loop (plans have no rate limit). Requests without an API key (signup, the MCP server's tool listing and find_endpoint) are limited per IP address instead. It also stops an account whose failed requests in the last hour cost far more than its successful ones paid for: failures are free within a limit (the message says so), and successes raise it. | Wait the seconds in the retry-after header, then retry. If your steady traffic needs more, ask [support](mailto:support@socialtap.dev) to raise the ceiling; without an API key, send one. |

### 451

451: 1 error code. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `unavailable_for_legal_reasons` | 451 | No | We removed this item after a copyright notice or another legal request, so we don't return it. The request was free. A list leaves such items out instead, with an items_omitted warning. | Do not retry. The item stays unavailable unless the notice is withdrawn. |

### 500

500 Internal Error: 1 error code. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `internal` | 500 | Yes, later | An unexpected error on SocialTap's side. | Retry with backoff. If it keeps happening, contact [support](mailto:support@socialtap.dev) with the request_id. |

### 502

502 Bad Gateway: 1 error code. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `upstream_error` | 502 | Yes, later | The platform failed, or answered with something we could not read. | Retry with backoff. Failed requests are free. |

### 503

503 Unavailable: 2 error codes. Free: errors cost nothing.

| Code | HTTP | Retry | Meaning | What to do |
| --- | --- | --- | --- | --- |
| `upstream_blocked` | 503 | Yes, later | The platform refused every attempt, our own retries included. | Retry later with backoff. Failed requests are free. |
| `overloaded` | 503 | Yes, later | SocialTap was at capacity, or paused by its operators, and did not start the request. Nothing was fetched or charged. | Retry after the seconds in the retry-after header. Failed requests are free. |

## Pagination

Items come in pages, **HN Search's ranking: the best match of the words first**. To get the next page, send **the same request plus `cursor`** set to `page.next_cursor` from the previous response. Keep every other parameter the same (limit, max_age and fields may change); anything else that differs is rejected. When `page.has_more` is `false` the walk is over, and `page.end_reason` says why: only `end` means you have every item (the others say the platform or a page cap stopped the walk early).

- **Order**: `sort=relevance` (the default): HN Search's ranking: the best match of the words first; `sort=newest`: newest first
- **Page size**: Up to 100 comments (the default; `limit` lowers it)
- **Cursor lifetime**: 24 hours; don't store cursors
- **Price**: Per page, 1 credit

### The `page` object

| Field | Type | Description |
| --- | --- | --- |
| `page.has_more` | `boolean` | True when another page exists; then next_cursor is set. An empty page with has_more true is not the end. |
| `page.next_cursor` | `string` or `null` | Pass as cursor, with the same other parameters, to get the next page; null at the end. |
| `page.next_cursor_expires_at` | `string` (date-time) or `null` | When next_cursor stops working, RFC 3339 UTC; null at the end. Don't store cursors. |
| `page.end_reason` | `string` or `null` | Why there are no more pages; null while has_more is true. end: the list ended, or reached since/until. platform_limit: the platform stops serving further pages although the list goes on. max_pages: SocialTap's cap on pages in one walk. New values may be added. A short or empty page is not the end: only has_more false is.<br>One of `end`, `platform_limit`, `max_pages` |

### Next page

cURL: next page:

```shell
curl -G 'https://socialtap.dev/v1/hackernews/search_comments' \
  --data-urlencode 'query=rust borrow checker' \
  --data-urlencode 'cursor=c1.EXAMPLE_CURSOR' \
  -H "x-api-key: $SOCIALTAP_API_KEY"
```

- The price is per page: each successful page costs the same, whatever `limit` is and however many items it has. Empty pages are charged too (an empty page with `has_more: true` is not the end).
- A `limit` below 100 doesn't save credits: the next request serves the rest of the same platform page, which is fetched again and charged as a page each time, and items can shift between those requests if the platform changes the list. Omitting `limit` is the cheapest way to walk the list.
- Errors are free, including a rejected cursor.
- Pages served from the cache are free.
- An expired or altered cursor, or one sent with different parameters, fails with [`invalid_cursor`](https://socialtap.dev/docs/errors/index.md#invalid_cursor) (HTTP 400): Repeat the original request without cursor to start the list again.

[Pagination guide](https://socialtap.dev/docs/pagination/index.md)

## Pricing

From **1 credit** per page. Charged per successful page, whatever its number of items. Errors are free. The price depends on the source that produced the result, reported as `meta.source`.

| Source | Price per successful page | Per 1,000 pages |
| --- | --- | --- |
| Platform API `api` | 1 credit | $0.75–$1.97 |

**Balance, sources and what you were charged**

Your balance must be at least **1 credit** to start a request; with less, it is refused with [`insufficient_credits`](https://socialtap.dev/docs/errors/index.md#insufficient_credits) (HTTP 402) and costs nothing. Only that minimum is set aside when the request starts.

| meta.source | Source | What it means |
| --- | --- | --- |
| `api` | Platform API | The platform's public, logged-out API. |

`meta.credits_charged` and the `x-credits-charged` header show what you paid. [How credits work](https://socialtap.dev/docs/credits/index.md) · [Packs](https://socialtap.dev/pricing/index.md)

Endpoint ID `hackernews.search_comments` · operationId `hackernews.search_comments.list` · [Product overview](https://socialtap.dev/apis/hackernews/search-comments/index.md)

---

[HTML version](https://socialtap.dev/docs/reference/hackernews/search-comments/) · [Docs index for AI agents](https://socialtap.dev/llms.txt) · [All docs in one file](https://socialtap.dev/llms-full.txt)
