Why does the skill require a reproduction before a fix?

A reproduction ensures the root cause is correctly identified and provides a baseline for the mandatory regression test, preventing future regressions.

How does it handle data durability issues?

It validates Write-Ahead Logs (WAL), manifest updates, and atomic renames to ensure the engine maintains integrity during crashes.

Can I use this for query performance issues?

Yes, the workflow includes observability steps using 'explain' and 'profile' flags to narrow down execution bottlenecks.

What is the output of this bug-hunting process?

The skill produces a reproduction artifact, fault localization, root cause analysis, a minimal fix plan, and a validated regression test.

Does this skill work with all Searchlite interfaces?

Yes, it includes specific debugging probes for the CLI, HTTP API, embedded Rust API, WASM/IndexedDB bindings, and the C FFI.

Searchlite Bug Hunting Workflow

Name: Searchlite Bug Hunting Workflow
Author: davidkelley

bydavidkelley

0•

보안 및 테스팅

Implements a structured, reproduction-first workflow for identifying and fixing bugs within the Searchlite search engine ecosystem.

This skill provides a rigorous framework for debugging the Searchlite search engine, covering its core indexing logic, CLI, HTTP API, and WASM bindings. It enforces a discovery-first approach that prioritizes system durability and data integrity, ensuring every fix is accompanied by a minimal reproduction case and a permanent regression test. It is particularly effective for investigating complex issues like intermittent crashes, data loss risks, and result inaccuracies without introducing unnecessary refactorings.

주요 기능

01Durability-focused checks for WAL, manifest atomicity, and fsync ordering

02Reproduction-first mandate to ensure bugs are verified before code modification

03Diagnostic probes for query execution modes including BM25, WAND, and BMW

040 GitHub stars

05Standardized discovery phase to map architecture and establish failure boundaries

06Automated integration with cargo-based quality suites and regression testing

사용 사례

01Investigating non-deterministic 'flaky' test failures or intermittent system crashes

02Diagnosing search correctness issues where results or aggregations appear inaccurate

03Verifying data durability and recovery after simulated system crashes or power failures

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add davidkelley/searchlite bug-hunting

For use in Claude.ai and ChatGPT

Download Skill