Repository navigation
feat: bound repository context size in load_repo_context (#24) - #45
rishiiicreates wants to merge 1 commit into
Conversation
|
Important
This repository does not receive automatic reviews because it has fewer than 10 stars. ⚙️ Run configuration
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
|
Hello 👋 This PR has had no activity for more than 2 weeks. If you are still working on it, please push an update or leave a comment. Ping a maintainer if you believe it is ready for review or merge! This PR will be automatically closed in 7 days if there is no further activity. |
|
This PR was automatically closed after being stale for 7 days with no activity. 😔 If you would like to continue, please reopen it and ping a maintainer for a review. Thank you for your contribution to AOSSIE! 🙏 |
Addressed Issues:
Fixes #24
Screenshots/Recordings:
N/A
Additional Notes:
• Introduced
_read_capped_filehelper withDEFAULT_MAX_FILE_CHARS = 2000andTRUNCATION_MARKER = "\n... (truncated)"to bound individual markdown files (operational data, architecture, core context, instructions, skills, and README).• Introduced
_append_bounded_sectionenforcingDEFAULT_MAX_TOTAL_CONTEXT_CHARS = 12000across all assembled sections, ensuring the joined repository context never exceeds the LLM prompt budget.• Replaced ad-hoc README truncation with the shared
_read_capped_filestrategy for consistency across all markdown context sources.• Preserved all existing intent-based file routing, keyword matching, and error logging behavior.
• Added dedicated test suite
scripts/test_bounded_context.pycovering capped reads, section budget bounds, mock repo bounds, and real repo context loading (23/23 assertions passing).• Verified existing
scripts/test_link_and_issue_features.py(56/56 passing) andscripts/routing_smoke_check.py.Checklist
Used AI assistance for brainstorming and structuring unit test cases in
scripts/test_bounded_context.py. All logic, budget calculations, and test suites were reviewed, executed, and verified locally with 100% passing results.