LingXiEngineering Workflow for Codex and Claude Code

Turn rough requests into engineer-ready tasks, challenge quality before implementation starts, and distill durable engineering judgment into reusable memory.

Get Started

GitHub

📝

Stronger Task Writing

Turn rough requests into engineer-ready task docs with clearer boundaries, constraints, acceptance criteria, and implementation guidance.

🔎

Sharper Vetting

Challenge task quality before implementation and surface ambiguity, hidden risk, weak acceptance coverage, and shaky framing early.

🧠

Engineering Taste Memory

Distill reusable engineering judgment from historical sessions so future work can inherit durable engineering choices.

⚙️

Small Surface, Strong Core

Keep the visible workflow narrow with task and vet while the memory layer continuously strengthens meaningful repository work underneath.

🧪

Semantic Judgment with Deterministic Safety

Use LLM reasoning for extraction, adjudication, governance, and ranking, with deterministic scripts for contracts, state safety, and persistence.

🔁

Continuous Background Distillation

Session-distill selects valid sessions, runs taste extraction and adjudication, governs memory writes, and lets generic meaningful repository turns receive the smallest useful memory brief through the repo-local hook (Codex or Claude Code).

Start with sharper work, then keep raising the bar

Install LingXi into a repository, use task to create engineer-ready tasks, use vet to challenge quality before implementation, and let session-distill accumulate durable engineering taste in the background.

Get Started GitHub