Groundhog: less déjà vu in CI

Description and use

Groundhog reduces repetitive effort in CI by grouping failures and retrieving prior resolutions automatically.

Problem

Recurring CI failures cause repeated effort to find solutions, even when the error wording changes.

Outcome

Groundhog reduces repeated effort by grouping CI failures and retrieving prior resolutions, achieving 70% correct suggestions in a stress test with 40 known variants.

Workflow

• Import a JUnit report into Groundhog
• Groundhog normalizes and groups identical symptoms
• The app encodes symptoms with MiniLM and retrieves matching historical notes
• Developer reviews the suggested resolution and evidence
• If correct, the resolution is saved as a note
• Future runs can retrieve the same note automatically

Evidence

Import a JUnit report, group repeated symptoms, and retrieve a resolution you recorded before, even when the error's wording changes.

Source

The gates were calibrated on 13 cases against three memories. The original held‑out set gave lexical and semantic retrieval identical known‑case ranking, so I expanded the test without retuning either policy.

Source

MiniLM ranked better here, but its suggestions were 64% correct, versus 70% for Jaccard.

Source