We study the WCAG compliancy and state of the metadata of PDF documents released under the Dutch Open Government Act (Woo). The results show that, in line with previous research on WCAG compliancy of PDF documents, only a fraction (0.2% of 31K) of the evaluated documents were WCAG compliant, with 160.407 WCAG related error instances in total. Five errors (out of 1.324) made up 68% of the total error instances, and 20 errors caused 95% of them. We have demonstrated that several of the errors in the top 20 can reliably be repaired with either existing Python packages or by using LLMs. We have automatically repaired six errors, reducing the total number of error instances in the dataset by 65K (40.5%). From the six defined essential metadata categories, the document language was least often missing in the documents (53.6% missing). Subject or description were most often missing, with a rate of 92.7%. Utilizing basic Python libraries and ChatGPT-4o for more complex metadata fields, our metadata field repairs had success rates between 69 and 92%. Repairing title metadata had a ROUGE-2 F1 score of .76.