@mrscraper/opencode 0.1.0 → 0.3.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +22 -0
- package/README.md +110 -68
- package/index.ts +73 -5
- package/package.json +3 -2
- package/skills/mrscraper/SKILL.md +64 -219
- package/skills/mrscraper-fetch/SKILL.md +97 -115
- package/skills/mrscraper-serp/SKILL.md +33 -40
- package/skills/mrscraper-scrape/SKILL.md +0 -206
|
@@ -1,35 +1,33 @@
|
|
|
1
1
|
---
|
|
2
2
|
name: mrscraper-fetch
|
|
3
3
|
description: |
|
|
4
|
-
Retrieve
|
|
4
|
+
Retrieve a known public URL with MrScraper's scraperAPI tool as HTML, Markdown, a screenshot, or a combination. Use for reading, summarizing, archiving, or analyzing a page and before local extraction; supports automatic browser fallback, real-device routing, proxies, and retries. Use mrscraper-serp when no URL is known.
|
|
5
5
|
---
|
|
6
6
|
|
|
7
|
-
#
|
|
7
|
+
# Retrieve Page Content with MrScraper MCP
|
|
8
8
|
|
|
9
|
-
Use the
|
|
10
|
-
|
|
11
|
-
|
|
12
|
-
handing the page to another LLM by default. Use
|
|
9
|
+
Use the scraperAPI tool supplied by the MrScraper MCP server whenever the user
|
|
10
|
+
already has a public URL. Keep using the retrieved content for summarization,
|
|
11
|
+
comparison, transformation, and structured output. Use
|
|
13
12
|
[mrscraper](../mrscraper/SKILL.md) for connection and authentication checks,
|
|
14
|
-
|
|
13
|
+
account status, or broader routing, and
|
|
14
|
+
[mrscraper-serp](../mrscraper-serp/SKILL.md) when discovery must happen first.
|
|
15
15
|
|
|
16
16
|
## Page Retrieval Context
|
|
17
17
|
|
|
18
|
-
|
|
18
|
+
scraperAPI calls MrScraper's
|
|
19
19
|
[page-fetching service](https://docs.mrscraper.com/docs/features/unblocker) once.
|
|
20
|
-
Browser rendering executes page JavaScript,
|
|
21
|
-
|
|
22
|
-
|
|
23
|
-
|
|
24
|
-
rendering.
|
|
20
|
+
Browser rendering executes page JavaScript, geographic routing selects the
|
|
21
|
+
proxy country, selector waits let delayed content appear, and homepage
|
|
22
|
+
navigation visits the site root before the target. Real-device routing (super)
|
|
23
|
+
is independent of browser rendering.
|
|
25
24
|
|
|
26
|
-
Use
|
|
27
|
-
with the site's requirements.
|
|
28
|
-
does not work with a basic request.
|
|
25
|
+
Use scraperAPI only for content the user is authorized to access and in
|
|
26
|
+
accordance with the site's requirements.
|
|
29
27
|
|
|
30
28
|
Start with the URL alone and add only the controls the target needs. Plan-token
|
|
31
29
|
usage is based on runtime and bandwidth: one token per 30 seconds and one token
|
|
32
|
-
per 0.
|
|
30
|
+
per 0.25 MB, rounded up per component. Resource blocking can reduce bandwidth
|
|
33
31
|
for text-focused pages. Retries stop when the request succeeds, max_retries is
|
|
34
32
|
reached, or the running token total reaches token_cap. The initial request
|
|
35
33
|
always runs even when it exceeds that cap. See the
|
|
@@ -40,96 +38,79 @@ maintained calculation.
|
|
|
40
38
|
|
|
41
39
|
Confirm the target URL and what the user wants:
|
|
42
40
|
|
|
43
|
-
- Preserve a raw source before extraction, transformation, or comparison.
|
|
44
41
|
- Read, summarize, cite, or inspect the page.
|
|
45
42
|
- Check whether specific text appears.
|
|
46
|
-
- Archive the
|
|
43
|
+
- Archive the page or capture a screenshot.
|
|
47
44
|
- Produce fields, JSON, tables, or other structured output with local logic.
|
|
48
|
-
- Verify or supplement a managed scrape or saved result.
|
|
49
45
|
- Load JavaScript-rendered or geo-sensitive content.
|
|
50
46
|
|
|
51
|
-
|
|
52
|
-
|
|
53
|
-
|
|
54
|
-
requests managed extraction, or after fetch-led exploration has produced a
|
|
55
|
-
stable output schema and managed extraction still offers a concrete benefit.
|
|
56
|
-
Use [mrscraper-serp](../mrscraper-serp/SKILL.md) when discovery must happen
|
|
57
|
-
first.
|
|
47
|
+
When many pages share a layout, retrieve representative pages, define one local
|
|
48
|
+
extractor, retrieve the remaining pages in parallel when safe and proportional,
|
|
49
|
+
and apply the same extractor to every saved response.
|
|
58
50
|
|
|
59
|
-
|
|
51
|
+
## Step 2 — Retrieve the Page
|
|
60
52
|
|
|
61
|
-
|
|
53
|
+
The client may namespace the tool name; select scraperAPI from the MrScraper
|
|
54
|
+
MCP provider. Start with one call containing only the required URL, which
|
|
55
|
+
requests HTML:
|
|
62
56
|
|
|
63
|
-
|
|
64
|
-
|
|
65
|
-
|
|
66
|
-
4. Apply the same local extractor to every saved response.
|
|
67
|
-
|
|
68
|
-
For roughly 100 same-layout pages, concurrent fetches followed by one local
|
|
69
|
-
batch extraction are often faster than 100 separate backend-LLM extractions.
|
|
70
|
-
This also preserves every raw input for later recovery or schema changes.
|
|
71
|
-
|
|
72
|
-
## Step 2 — Run the Fetch
|
|
57
|
+
{
|
|
58
|
+
"url": "https://www.scrapethissite.com/pages/simple/"
|
|
59
|
+
}
|
|
73
60
|
|
|
74
|
-
|
|
75
|
-
provider. Start with one call containing only the required URL:
|
|
61
|
+
Request Markdown instead of HTML for readable text:
|
|
76
62
|
|
|
77
63
|
{
|
|
78
|
-
"url": "https://www.scrapethissite.com/pages/simple/"
|
|
64
|
+
"url": "https://www.scrapethissite.com/pages/simple/",
|
|
65
|
+
"html": false,
|
|
66
|
+
"markdown": true
|
|
79
67
|
}
|
|
80
68
|
|
|
81
|
-
The result is available in MCP structuredContent and as formatted JSON text
|
|
69
|
+
The result is available in MCP structuredContent and as formatted JSON text.
|
|
70
|
+
The requested representations are in data:
|
|
82
71
|
|
|
83
72
|
{
|
|
84
73
|
"status_code": 200,
|
|
85
|
-
"data":
|
|
74
|
+
"data": {
|
|
75
|
+
"html": "<!DOCTYPE html>...",
|
|
76
|
+
"markdown": null,
|
|
77
|
+
"screenshot": null,
|
|
78
|
+
"token_usage": 3,
|
|
79
|
+
"runtime": 18.02,
|
|
80
|
+
"retry_count": 0,
|
|
81
|
+
"error": null
|
|
82
|
+
},
|
|
86
83
|
"headers": {
|
|
87
|
-
"
|
|
84
|
+
"x-status-code": "200"
|
|
88
85
|
}
|
|
89
86
|
}
|
|
90
87
|
|
|
91
|
-
Prefer structuredContent. Check isError, error,
|
|
92
|
-
|
|
88
|
+
Prefer structuredContent. Check isError, error, status_code, and
|
|
89
|
+
headers["x-status-code"] (the target page's status) before using data.
|
|
93
90
|
|
|
94
|
-
|
|
95
|
-
|
|
96
|
-
|
|
97
|
-
can return different content or failures for the same URL:
|
|
98
|
-
|
|
99
|
-
| browser_rendering | super_mode | Loading path |
|
|
100
|
-
| --- | --- | --- |
|
|
101
|
-
| false | false | Standard routing with the non-browser loader. |
|
|
102
|
-
| true | false | Standard routing with browser loading and JavaScript. |
|
|
103
|
-
| false | true | Real-device routing with the non-browser loader. |
|
|
104
|
-
| true | true | Real-device routing with browser loading and JavaScript. |
|
|
105
|
-
|
|
106
|
-
Use browser rendering when the page depends on JavaScript:
|
|
91
|
+
A screenshot is a base64 JPEG that can be large. Request it only when the user
|
|
92
|
+
needs an image, and save it to a file rather than reading it into the
|
|
93
|
+
conversation:
|
|
107
94
|
|
|
108
95
|
{
|
|
109
|
-
"url": "https://www.scrapethissite.com/pages/
|
|
110
|
-
"
|
|
96
|
+
"url": "https://www.scrapethissite.com/pages/simple/",
|
|
97
|
+
"html": false,
|
|
98
|
+
"screenshot": true
|
|
111
99
|
}
|
|
112
100
|
|
|
113
|
-
|
|
114
|
-
browser loading is unnecessary or returns a worse response:
|
|
101
|
+
## Step 3 — Choose Page-Loading Options
|
|
115
102
|
|
|
116
|
-
|
|
117
|
-
|
|
118
|
-
|
|
119
|
-
}
|
|
103
|
+
Omit browser_rendering to let the service choose: it falls back to a browser
|
|
104
|
+
automatically when needed. true forces a browser and false disables it.
|
|
105
|
+
screenshot=true always uses a browser.
|
|
120
106
|
|
|
121
|
-
Use
|
|
107
|
+
Use real-device routing when routing may be the problem:
|
|
122
108
|
|
|
123
109
|
{
|
|
124
|
-
"url": "https://www.scrapethissite.com/pages/
|
|
125
|
-
"
|
|
126
|
-
"super_mode": true
|
|
110
|
+
"url": "https://www.scrapethissite.com/pages/simple/",
|
|
111
|
+
"super": true
|
|
127
112
|
}
|
|
128
113
|
|
|
129
|
-
Browser rendering is not a strictly stronger mode. Some sites fail or return
|
|
130
|
-
worse content with browser_rendering=true but load successfully when it is
|
|
131
|
-
false. Super Mode does not enable browser rendering.
|
|
132
|
-
|
|
133
114
|
Wait for delayed content with a CSS selector:
|
|
134
115
|
|
|
135
116
|
{
|
|
@@ -142,21 +123,17 @@ Use geographic routing or homepage navigation when the target requires it:
|
|
|
142
123
|
|
|
143
124
|
{
|
|
144
125
|
"url": "https://www.scrapethissite.com/pages/simple/",
|
|
145
|
-
"browser_rendering": true,
|
|
146
126
|
"geo_code": "ID",
|
|
147
127
|
"home_page": true
|
|
148
128
|
}
|
|
149
129
|
|
|
150
|
-
|
|
151
|
-
|
|
152
|
-
Bound resource use for a browser-rendered page:
|
|
130
|
+
Bound resource use:
|
|
153
131
|
|
|
154
132
|
{
|
|
155
133
|
"url": "https://www.scrapethissite.com/pages/ajax-javascript/#2015",
|
|
156
|
-
"browser_rendering": true,
|
|
157
134
|
"block_resources": true,
|
|
158
|
-
"max_retries":
|
|
159
|
-
"token_cap":
|
|
135
|
+
"max_retries": 1,
|
|
136
|
+
"token_cap": 10,
|
|
160
137
|
"timeout": 60
|
|
161
138
|
}
|
|
162
139
|
|
|
@@ -164,50 +141,55 @@ Bound resource use for a browser-rendered page:
|
|
|
164
141
|
|
|
165
142
|
| MCP input | Default | Request mapping | Use |
|
|
166
143
|
| --- | --- | --- | --- |
|
|
167
|
-
| url | required |
|
|
168
|
-
|
|
|
169
|
-
|
|
|
170
|
-
|
|
|
171
|
-
|
|
|
172
|
-
|
|
|
173
|
-
|
|
|
174
|
-
|
|
|
175
|
-
|
|
|
176
|
-
|
|
|
144
|
+
| url | required | Body url | Absolute HTTP or HTTPS target URL. |
|
|
145
|
+
| html | true | Query html | Request HTML content. |
|
|
146
|
+
| markdown | false | Query markdown | Request Markdown content. |
|
|
147
|
+
| screenshot | false | Query screenshot | Capture a base64 JPEG; forces browser rendering. |
|
|
148
|
+
| browser_rendering | automatic | Query browserRendering | true forces and false disables browser rendering. |
|
|
149
|
+
| super | false | Query super | Real-device routing, independent of browser rendering. |
|
|
150
|
+
| geo_code | US | Query geoCode | Proxy country, such as US, ID, or SG. |
|
|
151
|
+
| proxy_country | omitted | Query proxyCountry | Compatibility country option. |
|
|
152
|
+
| use_proxy | true | Query useProxy | Set false to skip the service's configured proxy. |
|
|
153
|
+
| proxy | omitted | Query proxy | Custom proxy as username:password@host:port. |
|
|
154
|
+
| wait_for_selector | omitted | Query waitForSelector | CSS selector to wait for. |
|
|
155
|
+
| home_page | false | Query warmStart | Visit the site root before the target page. |
|
|
156
|
+
| block_resources | false | Query blockResources | Block images, CSS, and fonts. |
|
|
157
|
+
| bypass_proxy | false | Query bypassProxy | Enable the proxy-bypass behavior. |
|
|
158
|
+
| cookie_jar | omitted | Query cookieJar | Cookie configuration string. |
|
|
159
|
+
| headers | omitted | Query headers | Target page request headers as an object or JSON string. |
|
|
160
|
+
| max_retries | 3 | Body maxRetries | Maximum retries; zero disables retries. |
|
|
161
|
+
| token_cap | omitted | Body tokenCap | Running token total that stops further retries. |
|
|
162
|
+
| timeout | 120 | Query timeout | Page timeout from 1 to 300 seconds. |
|
|
163
|
+
| stream | omitted | Query stream | Ask the API to stream; the response is still returned once. |
|
|
164
|
+
| discard_token_secret | empty | Header x-discard-token-secret | Optional upstream header. |
|
|
177
165
|
|
|
178
166
|
Authentication is handled by the MCP client's OAuth 2.1 connection and is
|
|
179
167
|
never a tool input.
|
|
180
168
|
|
|
181
169
|
## Step 4 — Inspect and Retry Deliberately
|
|
182
170
|
|
|
183
|
-
Start with
|
|
184
|
-
|
|
185
|
-
content:
|
|
171
|
+
Start with the URL alone unless the task already establishes a requirement. If
|
|
172
|
+
the response fails, is blocked, incomplete, or missing dynamic content:
|
|
186
173
|
|
|
187
|
-
1. Inspect the
|
|
188
|
-
2. Change one axis at a time:
|
|
174
|
+
1. Inspect status_code, data.error, headers["x-status-code"], and the content.
|
|
175
|
+
2. Change one axis at a time: force or disable browser_rendering, or set super
|
|
189
176
|
when routing may be the problem.
|
|
190
|
-
3.
|
|
191
|
-
value with browser_rendering=false.
|
|
192
|
-
4. Try the remaining untested combinations when the response is still unusable.
|
|
193
|
-
5. Add wait_for_selector, geo_code, or home_page only when evidence shows that
|
|
177
|
+
3. Add wait_for_selector, geo_code, or home_page only when evidence shows that
|
|
194
178
|
the target requires it.
|
|
195
|
-
|
|
179
|
+
4. Stop after a usable response unless the user requests a comparison.
|
|
196
180
|
|
|
197
|
-
Do not repeat an identical
|
|
198
|
-
|
|
199
|
-
|
|
200
|
-
session.
|
|
181
|
+
Do not repeat an identical request. Treat wait_for_selector as a CSS selector,
|
|
182
|
+
not a duration. Browser rendering loads a page; it does not click controls,
|
|
183
|
+
submit forms, or provide an authenticated interactive browser session.
|
|
201
184
|
|
|
202
185
|
## Step 5 — Deliver the Result
|
|
203
186
|
|
|
204
187
|
Answer the user's request from data. Keep the full envelope when headers or
|
|
205
188
|
diagnostics matter. When the user requests an archive, save the successful
|
|
206
|
-
response or
|
|
207
|
-
capability;
|
|
208
|
-
|
|
209
|
-
Keep
|
|
210
|
-
tables, JSON transformations, and extraction scripts from
|
|
211
|
-
|
|
212
|
-
|
|
213
|
-
as exhaustive source content.
|
|
189
|
+
response or the requested field with the environment's local file-writing
|
|
190
|
+
capability; scraperAPI itself has no output-path input.
|
|
191
|
+
|
|
192
|
+
Keep retrieved content as the source of truth for later steps. Build
|
|
193
|
+
summaries, tables, JSON transformations, and extraction scripts from it, and
|
|
194
|
+
retrieve the page again when a follow-up needs details that an earlier
|
|
195
|
+
transformation did not preserve.
|
|
@@ -1,28 +1,31 @@
|
|
|
1
1
|
---
|
|
2
2
|
name: mrscraper-serp
|
|
3
3
|
description: |
|
|
4
|
-
Discover public pages through Google with MrScraper using a search query or Google search URL. Use when no target URL is known or the user asks to inspect Google results. After discovery,
|
|
4
|
+
Discover public pages through Google with MrScraper's webSearch tool using a search query or Google search URL. Use when no target URL is known or the user asks to inspect Google results. After discovery, retrieve selected pages with mrscraper-fetch.
|
|
5
5
|
---
|
|
6
6
|
|
|
7
|
-
# Discover Pages with MrScraper MCP
|
|
7
|
+
# Discover Pages with MrScraper MCP Web Search
|
|
8
8
|
|
|
9
|
-
Use the
|
|
10
|
-
a search query rather than a known target URL. Use
|
|
11
|
-
[mrscraper](../mrscraper/SKILL.md) for connection troubleshooting,
|
|
12
|
-
|
|
9
|
+
Use the webSearch tool supplied by the MrScraper MCP server when a task begins
|
|
10
|
+
with a search query rather than a known target URL. Use
|
|
11
|
+
[mrscraper](../mrscraper/SKILL.md) for connection troubleshooting, account
|
|
12
|
+
status, or broader routing.
|
|
13
13
|
|
|
14
14
|
## Step 1 — Build the Query
|
|
15
15
|
|
|
16
|
-
Include the important entity, attribute, location, or date in
|
|
16
|
+
Include the important entity, attribute, location, or date in url, and set
|
|
17
|
+
raw=false to receive parsed results:
|
|
17
18
|
|
|
18
19
|
{
|
|
19
|
-
"
|
|
20
|
+
"url": "web scraping best practices",
|
|
21
|
+
"raw": false
|
|
20
22
|
}
|
|
21
23
|
|
|
22
24
|
The input can also be a complete Google search URL:
|
|
23
25
|
|
|
24
26
|
{
|
|
25
|
-
"
|
|
27
|
+
"url": "https://www.google.com/search?q=web+scraping&gl=us&hl=en&start=20",
|
|
28
|
+
"raw": false
|
|
26
29
|
}
|
|
27
30
|
|
|
28
31
|
For a Google URL, the tool reads q, gl, hl, and start. Explicit region,
|
|
@@ -33,7 +36,8 @@ language, and page inputs take precedence.
|
|
|
33
36
|
Set both country and language when results must match a locale:
|
|
34
37
|
|
|
35
38
|
{
|
|
36
|
-
"
|
|
39
|
+
"url": "running shoes",
|
|
40
|
+
"raw": false,
|
|
37
41
|
"region": "id",
|
|
38
42
|
"language": "id",
|
|
39
43
|
"page": 2
|
|
@@ -44,25 +48,21 @@ the current page before requesting another.
|
|
|
44
48
|
|
|
45
49
|
## Step 3 — Choose the Output
|
|
46
50
|
|
|
47
|
-
|
|
48
|
-
|
|
49
|
-
|
|
50
|
-
{
|
|
51
|
-
"query_or_url": "iphone 17",
|
|
52
|
-
"format": "json"
|
|
53
|
-
}
|
|
51
|
+
raw defaults to true, which returns result-page HTML as the Python SDK does.
|
|
52
|
+
Set raw=false for parsed JSON; organic results are in data.data.organic, next
|
|
53
|
+
to sections such as people_also_ask and pagination.
|
|
54
54
|
|
|
55
55
|
Request result-page HTML when the page itself is needed:
|
|
56
56
|
|
|
57
57
|
{
|
|
58
|
-
"
|
|
59
|
-
"format": "html"
|
|
58
|
+
"url": "iphone 17"
|
|
60
59
|
}
|
|
61
60
|
|
|
62
61
|
Use JavaScript rendering for dynamic result features such as AI Overview:
|
|
63
62
|
|
|
64
63
|
{
|
|
65
|
-
"
|
|
64
|
+
"url": "what is web scraping",
|
|
65
|
+
"raw": false,
|
|
66
66
|
"render_js": true
|
|
67
67
|
}
|
|
68
68
|
|
|
@@ -70,14 +70,14 @@ Use JavaScript rendering for dynamic result features such as AI Overview:
|
|
|
70
70
|
|
|
71
71
|
| MCP input | Default | Request mapping | Use |
|
|
72
72
|
| --- | --- | --- | --- |
|
|
73
|
-
|
|
|
73
|
+
| url | required | Body query | Search query or complete Google search URL. |
|
|
74
|
+
| raw | true | Body format=html | Return result-page HTML; set false to use format. |
|
|
75
|
+
| format | json | Body format | json or html when raw=false. |
|
|
74
76
|
| region | URL value or omitted | Body region | Google result country, such as us or id. |
|
|
75
77
|
| language | URL value or omitted | Body language | Google result language, such as en or id. |
|
|
76
78
|
| page | URL value or omitted | Body page | Positive, one-based results page. |
|
|
77
|
-
|
|
|
78
|
-
|
|
|
79
|
-
| raw | false | Body format=html | Deprecated compatibility alias; prefer format=html. |
|
|
80
|
-
| client_timeout | 120 | MCP upstream deadline | Maximum wait in seconds; not sent in the SERP body. |
|
|
79
|
+
| render_js | false | Body renderJs | Render JavaScript-dependent result features. |
|
|
80
|
+
| timeout | 600 | MCP upstream deadline | Maximum wait in seconds; not sent in the search body. |
|
|
81
81
|
|
|
82
82
|
Authentication is handled by the MCP client's OAuth 2.1 connection and is
|
|
83
83
|
never a tool input.
|
|
@@ -88,25 +88,19 @@ The complete response envelope is available in MCP structuredContent and as
|
|
|
88
88
|
formatted JSON text. Prefer structuredContent. Check isError, error, and
|
|
89
89
|
status_code before using data.
|
|
90
90
|
|
|
91
|
-
For format=json, inspect data for the available titles, URLs, snippets, and
|
|
92
|
-
other result fields. For format=html, data contains result-page HTML.
|
|
93
|
-
|
|
94
91
|
When the user requests a search artifact, save the successful data or full
|
|
95
|
-
envelope with local file-writing capability;
|
|
92
|
+
envelope with local file-writing capability; webSearch has no output-path
|
|
93
|
+
input.
|
|
96
94
|
|
|
97
95
|
## Step 5 — Continue with Relevant URLs
|
|
98
96
|
|
|
99
|
-
Select only URLs relevant to the user's goal, then
|
|
100
|
-
|
|
101
|
-
|
|
102
|
-
each selected page before deriving answers, comparisons, or structured output.
|
|
103
|
-
- Add [mrscraper-scrape](../mrscraper-scrape/SKILL.md) only when the user
|
|
104
|
-
explicitly requests managed extraction or it still offers a clear benefit
|
|
105
|
-
after the page structure and output schema are understood.
|
|
97
|
+
Select only URLs relevant to the user's goal, then load
|
|
98
|
+
[mrscraper-fetch](../mrscraper-fetch/SKILL.md) to retrieve and inspect each
|
|
99
|
+
selected page before deriving answers, comparisons, or structured output.
|
|
106
100
|
|
|
107
101
|
Run independent follow-up URLs in parallel when the environment supports safe
|
|
108
102
|
parallel execution. Keep the number of pages proportional to the requested
|
|
109
|
-
coverage.
|
|
103
|
+
coverage. If the user only requested URL discovery, stop after the search.
|
|
110
104
|
|
|
111
105
|
## Step 6 — Handle Weak or Missing Results
|
|
112
106
|
|
|
@@ -114,10 +108,9 @@ coverage.
|
|
|
114
108
|
- Rewrite noisy queries around the important entity, attribute, and location.
|
|
115
109
|
- Request another result page only after checking the current page.
|
|
116
110
|
- Retry once with render_js=true when a dynamic result feature is required.
|
|
117
|
-
- Increase
|
|
111
|
+
- Increase timeout when the MCP request deadline is too short.
|
|
118
112
|
- Use [mrscraper](../mrscraper/SKILL.md) when the connection or authentication
|
|
119
113
|
fails.
|
|
120
114
|
|
|
121
|
-
|
|
122
|
-
[mrscraper-scrape](../mrscraper-scrape/SKILL.md) to discover URLs within one
|
|
115
|
+
webSearch is for public Google discovery, not for discovering URLs within one
|
|
123
116
|
known site.
|
|
@@ -1,206 +0,0 @@
|
|
|
1
|
-
---
|
|
2
|
-
name: mrscraper-scrape
|
|
3
|
-
description: |
|
|
4
|
-
Run MrScraper's general, listing, or map agents for managed structured extraction or bounded URL discovery within a known site. Use when managed output is explicitly requested or justified after source-page inspection; use mrscraper-fetch for initial acquisition of known pages and mrscraper-serp when no starting URL is known.
|
|
5
|
-
---
|
|
6
|
-
|
|
7
|
-
# Extract Structured Data with MrScraper MCP
|
|
8
|
-
|
|
9
|
-
Do not use scrape for the first exploration of a known page. Start with
|
|
10
|
-
[mrscraper-fetch](../mrscraper-fetch/SKILL.md), inspect the complete raw
|
|
11
|
-
response, and prefer local analysis or a reusable extractor. Use scrape when
|
|
12
|
-
the user explicitly wants MrScraper-managed extraction, or after the agent
|
|
13
|
-
understands the website and has defined a stable output schema. Use
|
|
14
|
-
[mrscraper](../mrscraper/SKILL.md) for connection troubleshooting, saved runs,
|
|
15
|
-
account status, or broader routing.
|
|
16
|
-
|
|
17
|
-
For general and listing, MrScraper retrieves page HTML and asks a backend LLM to
|
|
18
|
-
interpret it according to prompt. This adds model-processing time, can narrow
|
|
19
|
-
the result to the requested fields, and repeats model work across many pages.
|
|
20
|
-
|
|
21
|
-
Before using general or listing, confirm all of the following:
|
|
22
|
-
|
|
23
|
-
1. The target or representative pages have already been fetched and inspected.
|
|
24
|
-
2. The site structure and required output fields are understood.
|
|
25
|
-
3. A stable output schema has been defined.
|
|
26
|
-
4. Managed extraction still offers a concrete benefit over local code, or the
|
|
27
|
-
user explicitly requested it.
|
|
28
|
-
|
|
29
|
-
If these conditions are not met, return to fetch. Requiring JSON, a table, or
|
|
30
|
-
named fields is not by itself a reason to call scrape. For large same-layout
|
|
31
|
-
sets—even roughly 100 pages—safe concurrent fetches plus one reusable local
|
|
32
|
-
extractor are often faster and preserve every raw response. The map agent is
|
|
33
|
-
separate URL discovery functionality and does not require an extraction schema.
|
|
34
|
-
|
|
35
|
-
## Step 1 — Choose an Agent
|
|
36
|
-
|
|
37
|
-
| Agent | Use it for | Required input | Available controls |
|
|
38
|
-
| --- | --- | --- | --- |
|
|
39
|
-
| general | One detail page or one extraction task | prompt | proxy_country, schema_prompt |
|
|
40
|
-
| listing | Repeated records or paginated listings | prompt | proxy_country, max_pages, schema_prompt |
|
|
41
|
-
| map | Discovering URLs across a site | URL only | max_depth, max_pages, limit, include_patterns, exclude_patterns |
|
|
42
|
-
|
|
43
|
-
The default agent is general. For map, omit prompt, schema_prompt, and
|
|
44
|
-
proxy_country. For general and listing, omit map-only controls. max_pages is
|
|
45
|
-
accepted by listing and map but not general.
|
|
46
|
-
|
|
47
|
-
agent selects the extraction workflow. mode independently selects the backend
|
|
48
|
-
execution tier: Cheap or Super. Omit mode to preserve the backend default, and
|
|
49
|
-
select Super only when the extraction requires the stronger mode.
|
|
50
|
-
|
|
51
|
-
## Step 2 — Define the Extraction
|
|
52
|
-
|
|
53
|
-
The examples below assume the decision gate above has been satisfied. Write a
|
|
54
|
-
prompt from the already-understood page structure and output schema, preserve
|
|
55
|
-
source values, and do not ask the model to infer unavailable values.
|
|
56
|
-
|
|
57
|
-
Detail-page example:
|
|
58
|
-
|
|
59
|
-
{
|
|
60
|
-
"url": "https://www.scrapethissite.com/pages/simple/",
|
|
61
|
-
"agent": "general",
|
|
62
|
-
"mode": "Super",
|
|
63
|
-
"prompt": "Extract each country's name, capital, population, and area. Preserve source values and omit unavailable fields."
|
|
64
|
-
}
|
|
65
|
-
|
|
66
|
-
Repeated-listing example:
|
|
67
|
-
|
|
68
|
-
{
|
|
69
|
-
"url": "https://www.scrapethissite.com/pages/forms/?page_num=1",
|
|
70
|
-
"agent": "listing",
|
|
71
|
-
"prompt": "Extract each hockey team's name, year, wins, losses, and win percentage.",
|
|
72
|
-
"max_pages": 5
|
|
73
|
-
}
|
|
74
|
-
|
|
75
|
-
Site-map example:
|
|
76
|
-
|
|
77
|
-
{
|
|
78
|
-
"url": "https://www.scrapethissite.com/",
|
|
79
|
-
"agent": "map",
|
|
80
|
-
"max_depth": 2,
|
|
81
|
-
"max_pages": 50,
|
|
82
|
-
"limit": 1000,
|
|
83
|
-
"include_patterns": "/pages/"
|
|
84
|
-
}
|
|
85
|
-
|
|
86
|
-
## Step 3 — Add Shape Guidance When Useful
|
|
87
|
-
|
|
88
|
-
schema_prompt accepts a JSON object, not a file path. If the user supplies a
|
|
89
|
-
local JSON Schema file, read and parse it with local file tools, confirm its
|
|
90
|
-
root is an object, and pass that object:
|
|
91
|
-
|
|
92
|
-
{
|
|
93
|
-
"url": "https://www.scrapethissite.com/pages/simple/",
|
|
94
|
-
"agent": "general",
|
|
95
|
-
"prompt": "Extract every country's name, capital, population, and area.",
|
|
96
|
-
"schema_prompt": {
|
|
97
|
-
"type": "object",
|
|
98
|
-
"properties": {
|
|
99
|
-
"countries": {
|
|
100
|
-
"type": "array",
|
|
101
|
-
"items": {
|
|
102
|
-
"type": "object",
|
|
103
|
-
"properties": {
|
|
104
|
-
"name": { "type": "string" },
|
|
105
|
-
"capital": { "type": "string" },
|
|
106
|
-
"population": { "type": "string" },
|
|
107
|
-
"area": { "type": "string" }
|
|
108
|
-
}
|
|
109
|
-
}
|
|
110
|
-
}
|
|
111
|
-
}
|
|
112
|
-
}
|
|
113
|
-
}
|
|
114
|
-
|
|
115
|
-
The MCP server appends this object to the extraction prompt as best-effort
|
|
116
|
-
shape guidance. It does not validate the returned data against the schema.
|
|
117
|
-
Validate locally when strict compliance is required.
|
|
118
|
-
|
|
119
|
-
## Step 4 — Set Parameters
|
|
120
|
-
|
|
121
|
-
| MCP input | Default | Request mapping | Use |
|
|
122
|
-
| --- | --- | --- | --- |
|
|
123
|
-
| url | required | Body url | Absolute HTTP or HTTPS starting URL. |
|
|
124
|
-
| agent | general | Body agent | Select general, listing, or map. |
|
|
125
|
-
| mode | service default | Body mode | Select Cheap or Super execution without changing the agent. |
|
|
126
|
-
| prompt | required for general/listing | Body message | Describe the fields or records to extract. |
|
|
127
|
-
| proxy_country | omitted | Body proxyCountry | Route general/listing through a country. |
|
|
128
|
-
| max_pages | service default | Body maxPages | Bound listing or map pages. |
|
|
129
|
-
| max_depth | service default | Body maxDepth | Bound map crawl depth. |
|
|
130
|
-
| limit | service default | Body limit | Bound map URL results. |
|
|
131
|
-
| include_patterns | omitted | Body includePatterns | Restrict map results to matching URLs. |
|
|
132
|
-
| exclude_patterns | omitted | Body excludePatterns | Exclude matching map URLs. |
|
|
133
|
-
| schema_prompt | omitted | Appended to message | JSON Schema object for general/listing guidance. |
|
|
134
|
-
|
|
135
|
-
Authentication is handled by the MCP client's OAuth 2.1 connection. The scrape
|
|
136
|
-
tool does not accept a credential or output-path input. Choose the smallest
|
|
137
|
-
practical limits and omit optional values when the service default is
|
|
138
|
-
appropriate.
|
|
139
|
-
|
|
140
|
-
## Step 5 — Wait for Completion
|
|
141
|
-
|
|
142
|
-
General and map usually finish quickly. Listing is synchronous and can take
|
|
143
|
-
several minutes even for one page.
|
|
144
|
-
|
|
145
|
-
Before starting a listing call, tell the user it may take several minutes. Keep
|
|
146
|
-
waiting on the original MCP call within the environment's supported tool
|
|
147
|
-
timeout. Do not submit a duplicate scrape because the first call is quiet or
|
|
148
|
-
still running.
|
|
149
|
-
|
|
150
|
-
## Step 6 — Use the Output
|
|
151
|
-
|
|
152
|
-
The complete API envelope is returned in MCP structuredContent and as formatted
|
|
153
|
-
JSON text. Prefer structuredContent and check isError, error, and status_code
|
|
154
|
-
before trusting the response.
|
|
155
|
-
|
|
156
|
-
A successful run normally stores its extracted value at:
|
|
157
|
-
|
|
158
|
-
structuredContent.data.data.data
|
|
159
|
-
|
|
160
|
-
The saved scraper UUID is normally at:
|
|
161
|
-
|
|
162
|
-
structuredContent.data.data.scraperId
|
|
163
|
-
|
|
164
|
-
When the user asks for a file, save the extracted value only after the tool
|
|
165
|
-
succeeds. Use local file-writing capability and report the path. Preserve the
|
|
166
|
-
full envelope separately when the scraper ID, response headers, or diagnostics
|
|
167
|
-
also matter.
|
|
168
|
-
|
|
169
|
-
Post-process only when the user requests filtering, merging, normalization,
|
|
170
|
-
CSV, a table, or another deliverable. Extracted fields named token or password
|
|
171
|
-
may be legitimate source data and are intentionally preserved.
|
|
172
|
-
|
|
173
|
-
## Step 7 — Preserve the Reproducible Scraper
|
|
174
|
-
|
|
175
|
-
Every successful scrape creates a saved AI scraper configuration by default.
|
|
176
|
-
Report scraperId when it is available and the extraction may need to run again.
|
|
177
|
-
|
|
178
|
-
Use the MrScraper rerun tool to apply the saved configuration to the same or
|
|
179
|
-
another URL:
|
|
180
|
-
|
|
181
|
-
{
|
|
182
|
-
"target": "https://www.scrapethissite.com/pages/forms/?page_num=2",
|
|
183
|
-
"type": "ai",
|
|
184
|
-
"scraper_id": "SCRAPER_UUID"
|
|
185
|
-
}
|
|
186
|
-
|
|
187
|
-
Prefer rerun over recreating the scrape definition when the user wants a
|
|
188
|
-
repeatable extraction. Explain that this reproduces the saved configuration,
|
|
189
|
-
not necessarily identical values when the page or model behavior changes.
|
|
190
|
-
|
|
191
|
-
The rerun tool also handles dashboard-built manual workflows and asynchronous
|
|
192
|
-
bulk jobs. Follow the full rerun workflow in
|
|
193
|
-
[mrscraper](../mrscraper/SKILL.md), including the required acknowledgment
|
|
194
|
-
before any manual rerun.
|
|
195
|
-
|
|
196
|
-
## Step 8 — Handle Failures
|
|
197
|
-
|
|
198
|
-
- Check isError, error, and status_code before using or saving output.
|
|
199
|
-
- Tighten the prompt when fields are missing or grouped incorrectly.
|
|
200
|
-
- Confirm the selected agent accepts every supplied input.
|
|
201
|
-
- Validate locally when downstream code requires a strict schema.
|
|
202
|
-
- Use [mrscraper-fetch](../mrscraper-fetch/SKILL.md) for page reading.
|
|
203
|
-
- Use [mrscraper-serp](../mrscraper-serp/SKILL.md) when discovery must happen
|
|
204
|
-
first.
|
|
205
|
-
- Never invent replacement fields or fill missing values with unsupported
|
|
206
|
-
assumptions.
|