1
00:00:00,159 --> 00:00:05,319
So, um, brought to you by Jellypod AI,
today we are jumping straight into the fresh

2
00:00:05,380 --> 00:00:09,599
Claude Code version 2.1.247 release.

3
00:00:09,659 --> 00:00:13,679
And man, Anthropic just dropped a really
clever tool inside it!

4
00:00:13,750 --> 00:00:14,150
Oh yeah?

5
00:00:14,214 --> 00:00:16,950
What did they, uh, what did they roll out
this time, mate?

6
00:00:16,958 --> 00:00:21,358
It is a new command called /claude-api
cost optimize.

7
00:00:21,446 --> 00:00:26,078
Basically, if you have an existing
codebase using the Anthropic SDK,

8
00:00:26,124 --> 00:00:31,278
Claude will literally profile your API
spend and work through five specific levers

9
00:00:31,331 --> 00:00:32,398
to bring your bill down.

10
00:00:32,796 --> 00:00:33,417
Fair dinkum.

11
00:00:34,056 --> 00:00:38,596
Lord knows back in my early dev days I
nearly melted a client budget by pushing

12
00:00:38,656 --> 00:00:41,196
un-cached script loops at midnight.

13
00:00:41,356 --> 00:00:43,517
So what are the five levers it actually
looks at?

14
00:00:43,542 --> 00:00:48,822
It checks prompt caching, token hygiene,
batch processing, effort level tuning,

15
00:00:48,882 --> 00:00:50,022
and model selection.

16
00:00:50,182 --> 00:00:54,582
You just run /claude-api cost optimize
right inside your terminal with your code

17
00:00:54,646 --> 00:00:54,902
open.

18
00:00:55,276 --> 00:00:58,917
Right, so it inspects the actual code
where you call the API?

19
00:00:59,000 --> 00:00:59,960
Exactly!

20
00:01:00,040 --> 00:01:05,000
It scans your API call sites, detects
redundant un-cached context blocks,

21
00:01:05,053 --> 00:01:09,000
flags places where you are routing simple
tasks to expensive models,

22
00:01:09,140 --> 00:01:12,280
and helps you make measured, one by one
changes.

23
00:01:12,292 --> 00:01:14,452
Ah, measured changes, that is key!

24
00:01:14,542 --> 00:01:18,692
Because, you know, if you just go nuclear
with aggressive prompt caching without

25
00:01:18,736 --> 00:01:22,692
checking your prompt structure
compatibility first, stuff breaks.

26
00:01:22,905 --> 00:01:27,572
Or if you drop the effort level down, you
might see benchmark regressions on complex

27
00:01:27,665 --> 00:01:28,212
tasks.

28
00:01:28,447 --> 00:01:28,907
Totally.

29
00:01:29,247 --> 00:01:34,407
And with batch processing, it makes sure
you only shift tasks over if they do not

30
00:01:34,427 --> 00:01:35,708
need realtime responses.

31
00:01:36,287 --> 00:01:39,628
You test each change against your output
quality before moving to the next one.

32
00:01:39,667 --> 00:01:41,027
That is proper smart.

33
00:01:41,267 --> 00:01:46,547
What else landed in 2.1.247 besides the
cost profiler?

34
00:01:46,542 --> 00:01:48,542
There is a new SendFeedback tool!

35
00:01:48,654 --> 00:01:53,262
When something goes sideways in a session,
Claude can automatically draft a feedback

36
00:01:53,319 --> 00:01:57,902
report for you to review and send using
/feedback slash command.

37
00:01:58,096 --> 00:02:00,897
Oh, that saves you typing out a whole bug
report from scratch!

38
00:02:01,756 --> 00:02:03,556
Can you turn it off if you do not want
drafts?

39
00:02:03,583 --> 00:02:06,703
Yep, you can toggle it off with the
feedbackDrafts setting.

40
00:02:06,847 --> 00:02:12,063
Also, cross session peer messages from
subagents now collapse by default into a

41
00:02:12,156 --> 00:02:16,943
clean one line preview, and you just hit
Control O to expand the full body.

42
00:02:17,397 --> 00:02:18,958
Thank goodness for that!

43
00:02:18,977 --> 00:02:23,077
My terminal transcript used to get so
cluttered with background chatter.

44
00:02:23,117 --> 00:02:25,617
What about that little pull request badge
in the footer?

45
00:02:25,667 --> 00:02:26,307
Ah, right!

46
00:02:26,387 --> 00:02:30,627
The prompt footer PR badge now skips its
GitHub re-check on terminal refocus

47
00:02:30,689 --> 00:02:33,667
whenever the last check was done less than
sixty seconds ago.

48
00:02:33,880 --> 00:02:37,027
So no more lag spikes every time you tab
back into your terminal.

49
00:02:37,681 --> 00:02:39,961
Small tweaks, big difference.

50
00:02:40,081 --> 00:02:40,901
Beautiful stuff!