Run a 3-lane causal trace and deep-interview to crystallize requirements from evidence.
npx skills add https://github.com/xz1220/oh-my-kimi --skill deep-dive-xz1220
Please help me install this Agent Skill. Skill: deep-dive Source: https://github.com/xz1220/oh-my-kimi/tree/main/skills/deep-dive Command: npx skills add https://github.com/xz1220/oh-my-kimi --skill deep-dive-xz1220
Deep-dive helps you investigate an unclear or evidence-heavy problem by first tracing the underlying causes and then turning the findings into a precise requirements specification instead of guessing.
Invoke deep-dive with your target problem statement, for example: /skill:deep-dive "Why does the production DAG fail intermittently on the transformation step?"
High-intent search queries and answers about installing and using this skill.
Root cause analysis for intermittent bugs requires tracing parallel hypotheses and synthesizing evidence. Deep-dive runs a 3-lane causal trace to investigate unclear system behaviors, injecting trace findings into an interactive interview to crystallize precise requirements.
Yes, you can generate requirements specifications from exploratory investigations by detecting brownfield environments and reusing codebase context. The pipeline adapts its approach for existing codebases, turning causal trace findings into structured spec artifacts.
The best way to plan changes with high causal uncertainty is running a parallel 3-lane causal trace to map hypotheses. Deep-dive conducts a focused deep-interview afterwards, using evidence-structured outputs to generate a final spec with dedicated trace findings.
Multi-agent causal tracing for bug diagnosis works by running a parallel 3-lane pipeline to explore different hypotheses simultaneously. It requires interactive lane confirmation, persists state for resuming, and saves trace artifacts into deterministic paths for later review.
Yes, deep-dive works with Kimi CLI to guide root-cause investigation interactively. You invoke the skill with your target problem statement, and it separates data from instructions while injecting trace-derived context into the interview protocol.
You should not use an interactive interview pipeline for requirements generation when the root cause is already clear or causal uncertainty is low. Deep-dive is specifically designed for evidence-heavy problems where underlying causes need parallel hypothesis tracing before spec generation.
alexmarucciInvestigate technical problems through hypothesis formation and experimentation.
e2itsClassify upstream spec changes and cascade invalidate downstream artifacts.
rongaredeBridges Kimi CLI into Claude Code with session persistence and agent/context injection.
kwt00Orchestrate a two-stage pipeline for causal investigation and requirements crystallization.
