How do I get all bars in a matplotlib bar chart?

If you want all bars, just capture the output from the plotting method. Its a list containing the bars:

import numpy as np
import matplotlib as mpl
import matplotlib.pyplot as plt

fig, ax = plt.subplots()

x = np.arange(5)
y = np.random.rand(5)

bars = ax.bar(x, y, color='grey')    
bars[3].set_color('g')

If you do want all Rectangle object in the axes, but these can be more then just bars, use:

bars = [rect for rect in ax.get_children() if isinstance(rect, mpl.patches.Rectangle)]

Another option that might be useful to some people is to access ax.containers. You have to be a little careful though as if your plot contains other types of containers you'll get those back too. To get just the bar containers something like

from matplotlib.container import BarContainer
bars = [i for i in ax.containers if isinstance(i, BarContainer)]

This can be pretty powerful with a few tricks (taking inspiration from the accepted example).

import numpy as np
import matplotlib.pyplot as plt

fig, ax = plt.subplots()

x = np.arange(5)
y = np.random.rand(2, 5)

ax.bar(x, y[0], width=0.5)
ax.bar(x + 0.5, y[1], width=0.5)

for bar, color in zip(ax.containers, ("red", "green")):
    # plt.setp sets a property on all elements of the container
    plt.setp(bar, color=color)

will give you:

If you add some labels to your plots you can construct a dictionary of containers to access them by label

import numpy as np
import matplotlib.pyplot as plt

fig, ax = plt.subplots()

x = np.arange(5)
y = np.random.rand(2, 5)

ax.bar(x, y[0], width=0.5)
ax.bar(x + 0.5, y[1], width=0.5, label='my bars')

named_bars = {i.get_label(): i for i in ax.containers}
plt.setp(named_bars["my bars"], color="magenta")

will give you

Of course, you can still access an individual bar patch within a container e.g.

import numpy as np
import matplotlib.pyplot as plt

fig, ax = plt.subplots()

x = np.arange(5)
y = np.random.rand(2, 5)

ax.bar(x, y[0], width=0.5)
ax.bar(x + 0.5, y[1], width=0.5)

plt.setp(ax.containers[0], color="black")
plt.setp(ax.containers[1], color="grey")
ax.containers[0][3].set_color("red")