ManyCP is not affiliated with, endorsed by or sponsored by any directory, Docker, OpenAI, Anthropic or GitHub named here. Names are used to say which service a fact is about.
This is a measured look at 10 of the 26 MCP directories: how long pull requests from outside contributors took to merge on the GitHub lists that take a server by pull request. Measured . The other 16 have no figure here. A web form, a command or an automatic copy leaves no public record of when a submission went in and when it was listed, and the GitHub MCP Registry takes requests in a discussion thread rather than a pull request[4], so there is nothing to time. Goose is left out on purpose: its repository is the whole Goose project, not only the extension list[5], so its pull requests would not say how an extension goes.
Over each list's most recent outside merges (up to 50), the median ran from 6 h on awesome-mcp-list to 9.7 days on ToolSDK MCP Registry. That is how earlier pull requests went. It is not how yours will go.
Pull requests merged on each list's GitHub repo, timed from opened to merged, read from GitHub's REST pulls list. Bot accounts ([bot]), the repo's owners, members and collaborators, and pull requests from a branch inside the list's own repo are left out. Median and 25th to 75th percentile over the last 50 such merges; the last-30-days column is the newest point of the rolling median, which can differ when the last 50 came in one batch; rolling 30-day median by merge date, weekly; windows with fewer than 3 merges skipped. The data is GitHub's list of pull requests[1].
Time to merge, by directory
Measured . Merged pull requests from outside contributors, opened to merged. Each time links to the merged pull requests on GitHub.
Measured . The 12 months run to that date. Open pull requests were counted with GitHub's search[2]; each number links to the same search. Where a directory has fewer than 5 outside merges in the 12 months, the row says so instead of giving a median.
The same data over twelve months
Each panel is one directory: the median time to merge over the 30 days up to a date, one point a week, for the 12 months to . Every panel has the same scale, so a line higher in its box is a slower stretch on any of them.
Rolling 30-day median time to merge, by merge date, in the 12 months to 1 Oct 2026. Every panel uses the same scale: days, on a log scale, up to 90 days. A gap in a line is a stretch with fewer than 3 merges in the 30 days.
awesome-mcp-serversRolling 30-day median time to merge
awesome-mcp-servers: rolling 30-day median time to merge, by the day the window ends. Windows with fewer than 3 merges are left out.
Window ending
Rolling median
Merges in window
9 Oct 2025
6.3 days
30
16 Oct 2025
5.9 days
33
23 Oct 2025
5.7 days
38
30 Oct 2025
2.8 days
23
6 Nov 2025
2.2 days
21
13 Nov 2025
2.1 days
21
20 Nov 2025
1.4 days
10
27 Nov 2025
1.4 days
10
4 Dec 2025
20 h
8
11 Dec 2025
48 days
97
18 Dec 2025
41 days
108
25 Dec 2025
41 days
108
1 Jan 2026
41 days
106
8 Jan 2026
1.4 days
37
15 Jan 2026
1.3 days
35
22 Jan 2026
18 h
43
29 Jan 2026
17 days
163
5 Feb 2026
26 days
143
12 Feb 2026
27 days
133
19 Feb 2026
5.6 days
194
26 Feb 2026
12 h
99
5 Mar 2026
12 h
98
12 Mar 2026
23 h
225
19 Mar 2026
1.1 days
363
26 Mar 2026
2.4 days
418
2 Apr 2026
2.6 days
418
9 Apr 2026
2.4 days
401
16 Apr 2026
3.0 days
398
23 Apr 2026
2.7 days
263
30 Apr 2026
3.8 days
408
7 May 2026
4.1 days
355
14 May 2026
4.2 days
298
21 May 2026
4.8 days
231
28 May 2026
12 days
250
4 June 2026
16 days
189
11 June 2026
10 days
403
18 June 2026
8.7 days
492
25 June 2026
7.4 days
568
2 July 2026
5.2 days
394
9 July 2026
3.5 days
300
16 July 2026
7.0 days
324
23 July 2026
9.2 days
463
30 July 2026
8.9 days
563
6 Aug 2026
8.9 days
519
13 Aug 2026
8.6 days
349
20 Aug 2026
8.6 days
350
27 Aug 2026
4.5 days
76
3 Sept 2026
21 days
114
10 Sept 2026
19 days
493
17 Sept 2026
14 days
747
24 Sept 2026
11 days
935
1 Oct 2026
9.8 days
822
awesome-remote-mcp-serversRolling 30-day median time to merge
awesome-remote-mcp-servers: rolling 30-day median time to merge, by the day the window ends. Windows with fewer than 3 merges are left out.
Window ending
Rolling median
Merges in window
10 Sept 2026
2 h
109
17 Sept 2026
6 h
228
24 Sept 2026
18 h
357
1 Oct 2026
23 h
462
Cline MarketplaceRolling 30-day median time to merge
Cline Marketplace: no 30-day window had 3 or more merges in the 12 months to 1 Oct 2026.
Window ending
Rolling median
Merges in window
Docker MCP CatalogRolling 30-day median time to merge
Docker MCP Catalog: rolling 30-day median time to merge, by the day the window ends. Windows with fewer than 3 merges are left out.
Window ending
Rolling median
Merges in window
9 Oct 2025
1.3 days
40
16 Oct 2025
1.3 days
28
23 Oct 2025
1.5 days
12
30 Oct 2025
under 1 h
18
6 Nov 2025
under 1 h
32
13 Nov 2025
2 h
39
20 Nov 2025
2 h
37
27 Nov 2025
2 h
37
4 Dec 2025
14 days
13
11 Dec 2025
16 days
10
18 Dec 2025
1.7 days
6
25 Dec 2025
1 day
5
1 Jan 2026
1.3 days
4
8 Jan 2026
14 h
6
15 Jan 2026
21 h
3
22 Jan 2026
11 days
4
29 Jan 2026
4.6 days
6
5 Feb 2026
4.6 days
4
12 Feb 2026
8.0 days
3
19 Feb 2026
13 days
6
26 Feb 2026
19 days
5
5 Mar 2026
23 days
4
12 Mar 2026
23 days
4
9 Apr 2026
26 days
4
16 Apr 2026
26 days
4
23 Apr 2026
14 days
5
30 Apr 2026
14 days
5
7 May 2026
14 days
3
9 July 2026
7.0 days
4
16 July 2026
13 days
3
30 July 2026
7.2 days
3
TensorBlock awesome-mcp-serversRolling 30-day median time to merge
TensorBlock awesome-mcp-servers: rolling 30-day median time to merge, by the day the window ends. Windows with fewer than 3 merges are left out.
Window ending
Rolling median
Merges in window
9 Oct 2025
3.3 days
3
30 Oct 2025
6.3 days
3
11 Dec 2025
3 h
3
18 Dec 2025
3 h
3
25 Dec 2025
3 h
3
1 Jan 2026
3 h
5
8 Jan 2026
1 h
6
15 Jan 2026
under 1 h
6
22 Jan 2026
under 1 h
9
29 Jan 2026
under 1 h
9
5 Feb 2026
3 h
9
12 Feb 2026
6 h
10
19 Feb 2026
8 h
14
26 Feb 2026
8 h
27
5 Mar 2026
9 h
55
12 Mar 2026
9 h
64
19 Mar 2026
15 h
81
26 Mar 2026
1 day
81
2 Apr 2026
1.1 days
95
9 Apr 2026
1.8 days
88
16 Apr 2026
1.6 days
90
23 Apr 2026
1.3 days
77
30 Apr 2026
2.4 days
66
7 May 2026
2.7 days
65
14 May 2026
3.1 days
59
21 May 2026
4.3 days
60
28 May 2026
4.5 days
62
4 June 2026
4.8 days
63
11 June 2026
13 days
162
18 June 2026
6.4 days
186
25 June 2026
5.7 days
191
2 July 2026
4.6 days
186
9 July 2026
1.5 days
103
16 July 2026
1.5 days
105
23 July 2026
1.7 days
109
30 July 2026
1.7 days
126
6 Aug 2026
1.7 days
128
13 Aug 2026
1.8 days
126
20 Aug 2026
2.4 days
126
27 Aug 2026
2.4 days
151
3 Sept 2026
1.4 days
155
10 Sept 2026
15 h
189
17 Sept 2026
11 h
201
24 Sept 2026
9 h
217
1 Oct 2026
10 h
288
awesome-mcp-listRolling 30-day median time to merge
awesome-mcp-list: rolling 30-day median time to merge, by the day the window ends. Windows with fewer than 3 merges are left out.
Window ending
Rolling median
Merges in window
12 Mar 2026
15 days
27
19 Mar 2026
15 days
27
26 Mar 2026
15 days
27
2 Apr 2026
15 days
27
9 Apr 2026
15 days
27
17 Sept 2026
60 days
74
24 Sept 2026
13 days
106
1 Oct 2026
3.4 days
122
ToolSDK MCP RegistryRolling 30-day median time to merge
ToolSDK MCP Registry: rolling 30-day median time to merge, by the day the window ends. Windows with fewer than 3 merges are left out.
Window ending
Rolling median
Merges in window
26 Feb 2026
4.4 days
3
5 Mar 2026
3.0 days
11
12 Mar 2026
1.1 days
22
19 Mar 2026
19 h
28
26 Mar 2026
14 h
28
2 Apr 2026
16 h
27
9 Apr 2026
13 h
25
16 Apr 2026
12 h
24
23 Apr 2026
13 h
19
30 Apr 2026
16 h
26
7 May 2026
23 h
13
14 May 2026
2.1 days
10
21 May 2026
2.4 days
16
28 May 2026
3.8 days
12
4 June 2026
6.4 days
17
11 June 2026
6.2 days
16
18 June 2026
6.5 days
20
25 June 2026
6.5 days
20
2 July 2026
6.4 days
21
9 July 2026
6.3 days
12
30 July 2026
17 days
49
6 Aug 2026
18 days
48
13 Aug 2026
17 days
49
20 Aug 2026
17 days
49
27 Aug 2026
13 days
36
3 Sept 2026
13 days
36
10 Sept 2026
14 days
35
17 Sept 2026
12 days
61
24 Sept 2026
12 days
61
1 Oct 2026
9.7 days
26
best-of-mcp-serversRolling 30-day median time to merge
best-of-mcp-servers: no 30-day window had 3 or more merges in the 12 months to 1 Oct 2026.
Window ending
Rolling median
Merges in window
ToolHive registryRolling 30-day median time to merge
ToolHive registry: no 30-day window had 3 or more merges in the 12 months to 1 Oct 2026.
Window ending
Rolling median
Merges in window
Kilo MarketplaceRolling 30-day median time to merge
Kilo Marketplace: no 30-day window had 3 or more merges in the 12 months to 1 Oct 2026.
Window ending
Rolling median
Merges in window
How it was measured
The list. For each repo, GitHub's list of closed pull requests, newest update first, read page by page until the newest 50 outside merges are known[1]. Without a token GitHub allows 60 requests an hour[3], so the run takes hours. The script is scripts/measure-queues.ts in the ManyCP repository, and it writes one dated file that this page reads.
The time. The days between a pull request being opened and being merged. A pull request that was closed without merging is not timed. Its share is its own column.
Who counts as outside. Left out: accounts whose name ends in [bot]; people GitHub marks as the repo's owner, a member or a collaborator; and any pull request opened from a branch inside the list's own repo. The branch rule is there because a maintainer whose membership of the organisation is private can otherwise be counted as an outsider.
The headline. The median and the middle half (25th to 75th percentile) of the newest 50 outside merges, whatever their age, with the sample size and its date range in the table. A directory with fewer than 5 outside merges in the 12 months gets no median.
The chart. A rolling median over 30 days, by merge date, weekly for 52 weeks. A window with fewer than 3 merges has no point, and the line breaks there.
ManyCP's own pull requests. 1 of the 50 pull requests in the TensorBlock awesome-mcp-servers sample were opened from the dayyad account, the one ManyCP submits its own listings from. They are counted like any other outside pull request.
What the numbers do not say
A typical past time is not a promise, and nothing here says a pull request of yours will merge in that time, or at all.
Closed without merge covers every reason a pull request ends that way: duplicates, spam, and ones their authors withdrew. The figure does not tell them apart.
An open count is a snapshot of a queue, not a wait. A list can hold hundreds of open pull requests and merge some every day.
These are 10 lists on GitHub. The directories that take a form have no equivalent figure, so none of them is ranked here.
Every list in the table is a free fork and a pull request, and you can open it yourself from your own account. ManyCP does not make any review faster, and it does not change who merges. It helps when the same server goes to several lists and the entry has to stay the same in each. It costs $19 once per server, or from $9 a month[6]. The free checker shows where a repo is listed today.