trace

Orchestrate evidence-ranked competing hypotheses to explain observed behaviors.

Updated Aug 23, 2026
One-click install
npx skills add https://github.com/kamelmh/opencode --skill trace-kamelmh
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: trace
Source: https://github.com/kamelmh/opencode/tree/main/skills/trace
Command: npx skills add https://github.com/kamelmh/opencode --skill trace-kamelmh

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill helps explain why an observed behavior or result happened by turning ambiguity into competing hypotheses and evidence-backed conclusions.

Core Features & Use Cases

  • Evidence-ranked hypothesis tracing: preserve clear distinctions between the observation, hypotheses, evidence for/against, the current best explanation, and the critical unknown.
  • Team-mode parallel investigation: orchestrate built-in tracer lanes (default 3) to gather evidence for and against different explanations in parallel.
  • Falsification-first rigor: require top hypotheses to be challenged via predictions, disconfirmation rules, and a rebuttal round before finalizing rankings.
  • Discriminating next probe: produce the single highest-value next step that would collapse uncertainty fastest, not a generic fix loop.
  • Use cases: runtime regressions, performance/latency behavior, architecture and orchestration root-cause analysis, and science/experimental result tracing.

Quick Start

Use the trace skill to explain why an outcome occurred by running it with your key observation: take the observation you saw and ask the AI to trace it, for example, “/trace <observation to trace>”.

Frequently Asked Questions about trace

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I perform root cause analysis for an ambiguous runtime regression?▼

Root cause analysis for runtime regressions requires orchestrating competing hypotheses and evidence-backed conclusions. You preserve the observation, run parallel evidence collection, apply falsification, and output a ranked synthesis with a discriminating next probe.

What is evidence-ranked hypothesis testing for debugging?▼

Evidence-ranked hypothesis testing for debugging distinguishes observations, hypotheses, evidence for and against, and the critical unknown. It challenges top explanations via predictions and disconfirmation rules before finalizing causal rankings.

How do I trace why a performance or latency behavior occurred?▼

Performance tracing for latency behavior involves orchestrating parallel investigations across multiple tracer lanes to gather evidence for and against competing explanations. This team-mode approach preserves structure, performs rebuttal rounds, and outputs a ranked synthesis.

Can I use parallel investigation lanes for causal architecture postmortem analysis?▼

Parallel investigation lanes for causal architecture postmortem analysis use team-mode orchestration over a tracer agent to gather evidence in parallel. This applies falsification-first rigor to hypotheses and outputs the highest-value discriminating next probe.

When should I not use a falsification-first tracing approach for debugging?▼

A falsification-first tracing approach for debugging is not suited for generic fix loops or simple known errors. It is designed for ambiguous, evidence-heavy investigations where causal explanations require competing hypotheses and disconfirmation rules.

What's the best way to identify the critical unknown in a scientific or experimental trace?▼

Identifying the critical unknown in scientific tracing requires ranking competing hypotheses via falsification and rebuttal rounds. The output highlights the single highest-value discriminating probe that would collapse uncertainty fastest.