docs(atl): add tips for attachments and comment formatting#3
Conversation
Add two behavioral tips for LLMs working with Jira issues: 1. Proactively download attachments when image references are seen 2. Note that comments are plain text (Markdown only works in descriptions)
Summary of ChangesHello @fank, 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 introduces specific behavioral guidelines for Large Language Models (LLMs) interacting with Jira issues via the Highlights
Using Gemini Code AssistThe 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
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 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
|
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request adds two important tips for LLMs interacting with Jira via the atl CLI. The tips clarify how to handle image attachments and the limitations of Markdown formatting in comments. The changes are clear and valuable for improving LLM behavior. I've added one suggestion to make the attachment-handling instruction more self-contained and explicit for the LLM, which should improve its reliability.
| - Embedded images show as \`[Image: filename]\` placeholders | ||
| - Use \`atl issue attachment PROJ-1234 --list\` to see attachments | ||
|
|
||
| **Important**: When a Jira issue description contains image references (e.g., \`[Image: filename.png]\`), always download and inspect attachments to understand the full context. Visual information is often essential to understanding requirements. |
There was a problem hiding this comment.
The 'Important' tip about attachments is a great addition. To make it more actionable for the LLM and reduce the chance of it failing to connect information from different parts of the document, I suggest including the full command sequence for listing and downloading attachments directly within this tip. The PR description outlines this sequence perfectly.
**Important**: When a Jira issue description contains image references (e.g., `[Image: filename.png]`), always download and inspect attachments to understand the full context. First, list attachments with `atl issue attachment PROJ-1234 --list` to get the attachment ID, then download it using `atl issue attachment PROJ-1234 --download <id>`. Visual information is often essential to understanding requirements.
* ci: set explicit contents: read permissions on workflows Addresses CodeQL alerts #1, #3, #9 (actions/missing-workflow-permissions). Both workflows only need read access to the repo contents, so the minimal permissions block is set at the workflow level. * fix: use exact-match check for git SSH rewrite config value The installers were using .includes('https://github.com/') to check the output of `git config --get url.git@github.com:.insteadOf`. That output is either empty or exactly the value we set ourselves, so exact match is both more correct and silences CodeQL's js/incomplete-url-substring-sanitization rule. Addresses CodeQL alerts #5, #6, #7, #8 across atl, discord, esq, and n8n installers. * fix(grafanactl): redact secrets from runConfig error messages runConfig threw errors containing the full args joined — for `config set contexts.X.grafana.password <value>` that embedded the plaintext password in error.message, which was then logged by the outer catch handler. Add a redactSensitiveArgs helper that replaces any value whose preceding key matches .token / .password / .secret / .apikey with <redacted> before building the error message. Addresses CodeQL alert #4 (js/clear-text-logging, error severity).
2 participants
Summary
Add two behavioral tips for LLMs working with Jira issues via the
atlCLI:[Image: ...]placeholders are seenChanges
How LLMs behave with these tips
Attachment inspection
Without the tip: When an LLM reads a Jira issue and sees
[Image: screenshot.png]in the description, it might:With the tip: The LLM will:
[Image: ...]as a signal that visual information existsatl issue attachment PROJ-1234 --listto find attachmentsatl issue attachment PROJ-1234 --download <id>Comment formatting
Without the note: An LLM might:
**bold**,- lists, etc.)With the note: The LLM will:
Test plan