# Trace Decisions

> Maintains a decision log when explicitly asked to record choices, evidence, and unresolved work across a task; excludes hidden reasoning, raw transcripts, and unrelated private context.

- Skill: `jovanipink/trace-decisions` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx get-skill add jovanipink/trace-decisions`
- Raw SKILL.md: https://api.skillmd.com/api/skills/jovanipink/trace-decisions/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Productivity
- License: MIT
- Author: JovaniPink (https://skillmd.com/u/jovanipink)
- Updated: 2026-11-15
- Page: https://skillmd.com/skills/jovanipink/trace-decisions

---


# Decision Evidence Trace

Record public reasons and observable evidence for consequential task decisions.

## Workflow

1. Confirm the task boundary, audience, retention need, and whether the trace stays in chat or may be written to a named file.
2. Default to an in-chat trace. Require explicit file authority before creating or appending to a repository artifact.
3. Record only decision-relevant entries with `timestamp`, `phase`, `decision`, `public basis`, `evidence references`, `outcome`, and `supersedes`.
4. Link to stable evidence rather than copying logs, prompts, source files, or private conversations into the trace.
5. Append corrections as new entries. Do not silently rewrite earlier entries when the trace is designated append-only.
6. At checkpoints, reconcile open decisions, invalidated assumptions, deviations, and the evidence needed next.

## Knowledge lifecycle handoff

When the trace supports a `knowledge-contract.v1` lifecycle, map observable research notes to `note` objects and externally supportable assertions to `claim` objects. Preserve stable project-scoped IDs, evidence status, editorial status, visibility, limitations, source relationships, and correction links. A correction appends a linked replacement version; it never rewrites a designated append-only entry.

Do not turn private reasoning into a claim object or treat a trace entry as source authority. Route the complete research-to-publication sequence to `run-research-workflow` when that skill is installed.

## Boundaries

- Do not record hidden chain-of-thought, internal deliberation, raw model reasoning, or private scratch work.
- Do not include secrets, credentials, personal data, proprietary text, raw transcripts, or unrelated user activity.
- Do not create a surveillance or activity log. Record only the named task's material decisions.
- Do not treat a trace entry as proof that the referenced action succeeded; verify external state separately.

## Output

Lead with the answer. Use the shortest format that retains the required evidence and limits; expand when requested or needed. The items below describe required content, not mandatory headings.

Return the current entries, unresolved evidence needs, and the trace location if a file was explicitly authorized. If a knowledge-lifecycle handoff was requested, include proposed note or claim IDs and relationships separately from the trace.

## Example

Synthetic example; not an observed client result.

Input: Record the decision to keep the old endpoint until client migration completes.

Output: Decision: retain the endpoint. Status: approved for this task. Evidence needed for removal: current client migration completion; no removal authority is implied.

