fix broken output format(yaml) prompt

[woung717]

User description

Since the code_feedback field is a property of the YAML output, a colon (:) should follow it. Some LLM models might omit the colon, causing a parsing error.

• Another suggestion
  I observed sometimes that text output in relevant_line property has broken indentation. I think adding multiple line example with indentation could better indent output.

  relevant_line: |
    xxx
    xxx # added

---

PR Type

bug fix, enhancement

---

Description

• Fixed missing colon after code_feedback field to correct YAML format in pr_agent/settings/pr_reviewer_prompts.toml.
• Improved indentation for relevant_line property to ensure proper formatting.

---

Changes walkthrough 📝

	Relevant files
Bug fix	pr_reviewer_prompts.toml Fix YAML output format and improve indentation                      pr_agent/settings/pr_reviewer_prompts.toml Fixed missing colon after code_feedback field to correct YAML format. Improved indentation for relevant_line property to ensure proper formatting. +1/-1

---

💡 PR-Agent usage:
Comment /help on the PR to get a list of all available PR-Agent tools and their descriptions

[codiumai-pr-agent-pro]

PR Reviewer Guide 🔍

⏱️ Estimated effort to review: 1 🔵⚪⚪⚪⚪

🏅 Score: 95

🧪 No relevant tests

🔒 No security concerns identified

🔀 No multiple PR themes

⚡ No key issues to review

[codiumai-pr-agent-pro]

PR Code Suggestions ✨

Category	Suggestion	Score
Possible issue	Correct the formatting of the 'code_feedback' block to ensure proper parsing Ensure that the 'code_feedback' block is properly formatted as a list in the TOML configuration. The current format might cause parsing issues. Use double quotes for string values and ensure proper indentation. pr_agent/settings/pr_reviewer_prompts.toml [168-171] code_feedback: -- relevant_file: | - directory/xxx.py - language: | + - relevant_file: "directory/xxx.py" + language: "toml" Apply this suggestion Suggestion importance[1-10]: 8 Why: The suggestion correctly identifies a potential formatting issue in the TOML configuration and provides a clear improvement to ensure proper parsing. Proper indentation and quoting string values are important for maintaining the integrity of the configuration file.	8

[KennyDizi]

I think it would be great if we moved the content to the right side a bit, like this

[mrT23]

thanks @woung717
This code is not on by default, and is not a major feature. improve should be used instead. So I don't see a big reason to further optimize this prompt. improve generates a valid YAML in a stable and consistent way with GPT4 and Claude

@KennyDizi output (and input) should be short and concise as possible, if it contains the same data. This saves time and money (fewer tokens).
Adding extra space is against this principle

[KennyDizi]

thanks @woung717 This code is not on by default, and is not a major feature. improve should be used instead. So I don't see a big reason to further optimize this prompt. improve generates a valid YAML in a stable and consistent way with GPT4 and Claude

@KennyDizi output (and input) should be short and concise as possible, if it contains the same data. This saves time and money (fewer tokens). Adding extra space is against this principle

That's good to know. If that's so, with the same approach, should we remove the space in the other similar places? @mrT23

[woung717]

I think the short and concise prompt principle is good. But modifying the prompt requires a fair amount of testing to make sure it keeps valid format, especially since this project parses structured text (YAML) directly from the LLM response. Maybe it is good to see if it is done correctly, but I think there is not much benefit for the effort. This is just my 2 cent.