Update Jules Triage Prompt

[milhy545]

This PR updates the system prompt for the Jules triage agent (src/mycoder/triage_agent.py) based on new directives.

The following changes were made:

1. Prompt Update: The JULES_SYSTEM_PROMPT was replaced with the new version provided by the user.
2. Format Adaptation: The prompt was adapted to use Python str.format() syntax (e.g., {available_labels} instead of !{echo $AVAILABLE_LABELS}) and JSON braces were escaped ({{/}}).
3. Safety Sanitization: The "Final Command Construction" instruction (which asked the LLM to generate a shell command) was removed. This is crucial because:
   • The triage_agent.py script expects and parses raw JSON output.
   • Including instructions to output shell commands leads to json.JSONDecodeError and crashes the agent.
   • It violates the "No Command Injection" safety constraint and memory instructions regarding prompt sanitization.
4. Constraint Update: Updated Constraint 4 to "Do not use command substitution $() in generated shell commands" as requested, although the prompt is still sanitized to output JSON.

The changes were verified using existing unit tests (tests/unit/test_triage_agent.py), confirming that prompt construction and JSON parsing remain functional.

---

PR created automatically by Jules for task 10651564694019365064 started by @milhy545

[google-labs-jules]

👋 Jules, reporting for duty! I'm here to lend a hand with this pull request.

When you start a review, I'll add a 👀 emoji to each comment to let you know I've read it. I'll focus on feedback directed at me and will do my best to stay out of conversations between you and other bots or reviewers to keep the noise down.

I'll push a commit with your requested changes shortly after. Please note there might be a delay between these steps, but rest assured I'm on the job!

For more direct control, you can switch me to Reactive Mode. When this mode is on, I will only act on comments where you specifically mention me with @jules. You can find this option in the Pull Request section of your global Jules UI settings. You can always switch back!

New to Jules? Learn more at jules.google/docs.

---

For security, I will only act on instructions from the user who triggered this task.

[github-actions]

🤖 Hi @milhy545, I've received your request, and I'm working on it now! You can track my progress in the logs for more details.

[gemini-code-assist]

Summary of Changes

Hello @milhy545, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request focuses on enhancing the robustness and safety of the Jules triage agent by updating its core system prompt. The primary goal is to ensure the agent consistently produces valid JSON output, preventing runtime errors and mitigating potential command injection vulnerabilities. These changes streamline the agent's operation and improve its reliability in processing triage tasks.

Highlights

• Prompt Update: The JULES_SYSTEM_PROMPT was replaced with a new version, incorporating updated directives.
• Format Adaptation: The new prompt was adapted to use Python str.format() syntax (e.g., {available_labels}) and JSON braces were properly escaped ({{/}}).
• Safety Sanitization: Instructions for the LLM to generate shell commands were removed to prevent json.JSONDecodeError and adhere to 'No Command Injection' safety constraints.
• Constraint Wording Update: Constraint 4 was updated to clarify 'Do not use command substitution $() in generated shell commands', reinforcing the safety measures.

Changelog

• src/mycoder/triage_agent.py
  • Updated the wording of Constraint 4 from 'Do not use command substitution $() in generated output' to 'Do not use command substitution $() in generated shell commands' for clarity and precision.
  • Modified the output specification instruction from 'Write a JSON array to the output.' to 'Write a JSON array to the output file.'
  • The JULES_SYSTEM_PROMPT was updated to a new version, which includes adapting to Python str.format() syntax, escaping JSON braces, and critically, removing instructions that could lead to the generation of shell commands.

Activity

• PR created automatically by Jules for task [10651564694019365064] started by @milhy545.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request updates the system prompt for the Jules triage agent. However, it introduces a security regression by narrowing a critical safety constraint against command substitution, potentially allowing malicious substitutions via prompt injection from untrusted GitHub issue content. Additionally, one of the changes introduces a contradiction in the prompt's instructions, which could confuse the language model.

[gemini-code-assist]

The change from 'Do not use command substitution $() in generated output' to 'Do not use command substitution $() in generated shell commands' significantly narrows the scope of protection against command injection. This creates a security vulnerability where an attacker could use prompt injection from untrusted GitHub issue content to trick the LLM into including malicious command substitutions in other parts of the JSON output (e.g., in the explanation field). This also re-introduces the term 'shell commands', creating a contradiction that could confuse the LLM, as the prompt is designed for strictly JSON output. The broader constraint is crucial for preventing arbitrary command execution if the output is processed by a shell-based system.

Suggested change

	4. **No Command Injection:** Do not use command substitution `$()` in generated shell commands.
	4. **No Command Injection:** Do not use command substitution $() in any part of the generated output.