Skip to main content

dev-trace

Evidence-driven causal tracing with competing hypotheses. Use when an observation has multiple plausible explanations and you need to systematically rank them with evidence FOR and AGAINST each. Pairs with @trail-tracer.

Source facts

Repository
evolution-foundation/evo-nexus
Last source activity
April 10, 2026 at 16:24
Detected SKILL.md language
English
Stars
552
Forks
181

Install options

The review-first prompt is selected by default. You can switch to a direct command or download a local copy.

Review the source files

Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.

Showing SKILL.md

SKILL.md
Source instructions · Read-only preview
name
dev-trace
description
Evidence-driven causal tracing with competing hypotheses. Use when an observation has multiple plausible explanations and you need to systematically rank them with evidence FOR and AGAINST each. Pairs with @trail-tracer.
# Dev Trace Derived from oh-my-claudecode (MIT, Yeachan Heo). Adapted for the EvoNexus Engineering Layer. Causal investigation with multi-hypothesis discipline. Generate at least 2 explanations, collect evidence FOR and AGAINST, rank by strength, end with a discriminating probe. ## Use When - Observation has multiple plausible causes - "Why did this happen?" — and you don't want to jump to the first answer - Multi-causal problem (intermittent bugs, regional anomalies, performance puzzles) - Need to preserve uncertainty rather than collapse to one answer prematurely ## Do Not Use When - Cause is obvious → just fix it via `@hawk-debugger` - Single-cause bug with clear stack trace → `@hawk-debugger` - Just needs codebase search → `@scout-explorer` ## Workflow Delegate to `@trail-tracer`: 1. **OBSERVE** — restate the observation precisely, no interpretation 2. **FRAME** — define the exact "why" question 3. **HYPOTHESIZE** — generate 2+ competing explanations 4. **GATHER EVIDENCE** — for each, collect evidence FOR and AGAINST in parallel 5. **APPLY LENSES** — systems / premortem / science 6. **REBUT** — strongest alternative challenges the leader 7. **RANK** — down-rank contradicted, weak-assumption, failed-prediction explanations 8. **SYNTHESIZE** — best explanation, explicitly provisional if needed 9. **PROBE** — name the critical unknown and the discriminating next experiment ## Output Saved to `workspace/development/debug/[C]trace-{topic}-{date}.md` using the `dev-trace-report.md` template. ## Pairs With - `@trail-tracer` (the executor) - `@hawk-debugger` (when trace converges to a clear bug) - `@apex-architect` (when trace reveals architectural issues)
View on GitHub