Traverse legacy codebases to capture facts and produce a spike document.
npx skills add https://github.com/gustavofsantos/dot-files --skill dead-reckoning
Please help me install this Agent Skill. Skill: dead-reckoning Source: https://github.com/gustavofsantos/dot-files/tree/main/skills/dead-reckoning Command: npx skills add https://github.com/gustavofsantos/dot-files --skill dead-reckoning
Structured analysis partner for complex legacy codebases. Use when the user wants to understand existing behavior, trace a call tree, investigate a bug, or answer a specific architectural question in a system that is hard to read (Clojure, verbose Java, Kotlin monorepos, or any poorly structured codebase). This skill drives the traversal session — enforcing a central question, managing a living knowledge base of facts and spikes, producing explicit coverage evidence, and finalizing a spike document at the end. The human validates; the agent traverses and translates. Activate on phrases like "analisa esse código", "quero entender como funciona", "como isso está implementado", "me ajuda a traçar", "trace this call", "how is this built", "what does X do in production", or any request to investigate existing code before making a decision or raising a PR. Do not use for greenfield design or planning without a codebase — use thinking-partner for that.
State a central question about the codebase and initiate the traversal session.
High-intent search queries and answers about installing and using this skill.
Legacy code analysis works by driving a structured traversal of the codebase around a central question. It captures facts and spikes during the traversal, producing explicit coverage evidence and a final spike document with validated findings.
Yes, analyzing poorly structured Java monorepos is supported. The system enforces a central question, loads knowledge from a spike and fact management system, and uses human validation to translate traversal results into engineering decisions.
The best way to document findings when investigating a bug in unfamiliar architectures is to manage a living knowledge base of facts and spikes. This workflow produces a final spike document that captures findings, decisions, and actionable next steps.
Yes, an existing codebase is required to perform structured code analysis. This traversal session is designed to diagnose bugs and answer architectural questions in current systems, and should not be used for greenfield design or planning without a codebase.
You should not use structured code traversal for greenfield design or planning without a codebase. It is specifically built for investigating existing code, tracing calls, and answering architectural questions in complex, verbose, or poorly structured legacy systems.
Fr-e-dAnalyze project directory structures to generate architectural summaries.
troykellyDocuments unknowns, creates research spikes, and forms resumable architecture for large projects.
papeete-consulting-labCapture code-derived decisions as structured candidate knowledge records.
twkevinzhangSynthesize research, codebase analysis, and requirements into architectural blueprints.
