@sipoon

sipoon-grill-me systematically validates task execution plans via 6w2h questioning, risk assessment, and acceptance criteria before work begins, reducing rework through structured pre-execution confirmation.

sipoon-diff-impact analyzes code change ripple effects through knowledge graph traversal to identify direct and indirect dependents before commit, but relies on pre-built graph indices and lacks concrete recovery strategies for missing or stale data.

skill-compounding is a meta-skill for identifying and extracting reusable solutions into formalized team knowledge artifacts, with a four-phase lifecycle from discovery through iteration to organizational propagation.

sipoon-lifecycle-hooks provides a heartbeat-polling system to simulate agent lifecycle events within openclaw constraints, implementing 5 of 13 hooks via sessions callbacks, cron tasks, and behavior detection, with 3 hooks explicitly unavailable due to missing kernel interception.

sqlite-backed entity-relation graph for skill compounding persistence, exposing atomic operations (add_entity, add_relation, query, traverse, get_path) with flexible node/edge schemas and soft integration points to skill-compounding workflows.

sipoon-codegraph-index builds pre-indexed symbol and call graphs using tree-sitter to accelerate code exploration in large repositories, reducing token consumption and tool invocations through structured queries instead of full-text scanning.

自动追问澄清模糊需求，分步确认设计方案，确保代码实现前需求明确，降低返工风险，支持分块确认与TDD流程。

定期扫描代码库架构质量，识别循环依赖、死代码、过大模块和高圈复杂度，生成报告指导重构优先级。