Skip to content

LoreCrawler: Dynamic retrieval of lore kernel reviews #77

@aditya-qti

Description

@aditya-qti

Is your feature request related to a problem? Please describe.
Traditional AI code reviewers often provide generic feedback that doesn't align with specific maintainer preferences or project conventions. Kernel maintainers have unique review styles, focus areas, and recurring concerns that are difficult to capture in static prompts.

Describe the solution you'd like
Retrieve relevant review examples from lore kernel and add them to the AiCodeReview prompt.

10-20 examples can be provided.

Note: Sometimes, review comments from a previous or the current series of the same patches can be retrieved - this
is a happy coincidence. The AI should be instructed not to use any author/reviewer names from these.

Design choices to consider
One-off call vs Agent tool with search flexibility or both!

If a lore search tool is being added, consider making tool_choice="required" for one of the initial iterations with this tool as the only one available.

Tests
Consider adding tests in tests/test_tools.py if a new agent tool is added.

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request
    No fields configured for Feature.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions