merge values in 2 columns based on duplicated value and date in other column

I have a dataframe consisting of 4 columns, the values in the first column occur exactly 2 times throughout the dataframe. The second column is a date and the 3th and 4th column are booleans. What I would like to accomplish, is to merge the duplicates and merge the value in the 3th and 4th column based on the date. See my example below to clarify things;

keys <- c("A","A","B","B","C","C")
dates <- as.POSIXct(c("2021-01-01", "2021-02-02","2021-03-03", "2021-04-04", "2021-05-05", "2021-06-06"))
vals1 <- c("T","F","T","F","F","F")
vals2 <- c("F","T","T","T","F","T")

df <- data.frame(key = keys,
                 date = dates,
                 val1 = vals1,
                 val2 = vals2)

This will produce the following dataframe

> df
  key       date val1 val2
1   A 2021-01-01    T    F
2   A 2021-02-02    F    T
3   B 2021-03-03    T    T
4   B 2021-04-04    F    T
5   C 2021-05-05    F    F
6   C 2021-06-06    F    T

Now I want to create a dataframe with 3 rows; 1 for each key value where val1 is the val1 value where the date is the earliest, and a val 2 value where the date is the largest. I have a piece of code that produces such outcome;

for (i in 1:nrow(df)){
  val1 <- df %>%
    filter(key == df$key[i]) %>%
    slice(which.min(date)) %>%
    ungroup() %>%
    select(val1)
  val2 <- df %>%
    filter(key == df$key[i]) %>%
    slice(which.max(date)) %>%
    ungroup() %>%
    select(val2)
  
  df$val1[df$key == df$key[i]] <- val1$val1
  df$val2[df$key == df$key[i]] <- val2$val2
}

df <- df[!duplicated(df$key),c("key","val1","val2")]

> df
  key val1 val2
1   A    T    T
3   B    T    T
5   C    F    T

Note in the final result, I am not interested in the date, only in the val1 and val2 values. The date is just there to determine which val1 or val2 value should be taken in the result. This code works fine for this dataframe, but I want to apply this on a large dataframe (~40k rows).

This is a problem, because I think this is a very inefficient way to do this, applying it on such a large dataframe takes forever to compute and my dataframes will only get larger in the future. I suspect it is due the for loop, but I cannot think of another way to solve this problem yet, therefore my question Hope someone can help me out, all help is greatly appreciated.

Kind regards

df %>% 
  group_by(key) %>% 
  summarise(val1 = val1[which.min(date)], 
            val2 = val2[which.max(date)]) %>%
  ungroup()

  key   val1  val2 
1 A     T     T    
2 B     T     T    
3 C     F     T   

Here is an alternative approach:

library(dplyr) # dplyr 1.1.0
df %>%
  summarise(val1 = val1[date == min(date)],
            val2 = val2[date == max(date)], .by=key)

  key val1 val2
1   A    T    T
2   B    T    T
3   C    F    T