How can I use python to change css attributes of an html document?

You could use regular expressions to make the necessary replacements as follows:

import re

test = """
      .org-link {
        /* org-link */
        color: #b58900;
        font-weight: bold;
        text-decoration: underline;
      }
"""

def fix(org_link):
    new_color = re.sub(r'(.*?color\s*?:\s*?)(.*?)(;)', r'\1#777\3', org_link.group(0), flags=re.S)
    return re.sub(r'(.*?)(\s+?text-decoration: underline;)(.*?)', r'\1\3', new_color, flags=re.S)

print re.sub(r'(org-link\s+\{.*\})', fix, test, flags=re.S)

This would convert the text as follows:

  .org-link {
    /* org-link */
    color:#777;
    font-weight: bold;
  }

It works by first identifying suitable org-link blocks and then first replacing the color and then removing any text-decoration entries.

The script could then be extended to carry this out on all of the HTML files in a given folder as follows:

import re
import glob

def fix(org_link):
    new_color = re.sub(r'(.*?color\s*?:\s*?)(.*?)(;)', r'\1#777\3', org_link.group(0), flags=re.S)
    return re.sub(r'(.*?)(\s+?text-decoration: underline;)(.*?)', r'\1\3', new_color, flags=re.S)

for html_file in glob.glob('*.html'):
    print html_file
    with open(html_file) as f_input:
        html = re.sub(r'(org-link\s+\{.*\})', fix, f_input.read(), flags=re.S)

    with open(html_file, 'w') as f_output:
        f_output.write(html)

Tested using Python 2.7.9