How do I install the Show Me Your Work skill?

Run npx skillmds@latest add canepro/show-me-your-work in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Show Me Your Work safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Show Me Your Work?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Show Me Your Work free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Show Me Your Work?

Canepro (@canepro) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Show Me Your Work by canepro — Agent Skill

Name: Show Me Your Work
Author: Canepro

Show Me Your Work

Keep a compact, reviewable decision ledger for long-running or unattended work. Use when a reviewer needs to reconstruct consequential choices and their evidence without reading private transcripts or replaying the entire task.

Canepro Updated Sep 21, 2026

File contents

Show me your work

Keep one decision ledger for the current task when a human will review the work after the fact. Record choices that changed scope, design, safety, verification, or disposition. Do not turn the ledger into a second transcript.

This workflow adapts pstack's show-me-your-work without transcript mining. Never read unrelated chats, raw session logs, browser history, or private records to reconstruct the ledger.

Ledger format

Use a TSV file with this header:

ts\tphase\tdecision\twhy\tevidence\tresult

Each cell stays on one line. Prefix spreadsheet-formula starters (=, +, -, or @) with a single quote when content is not fully controlled.

ts: ISO 8601 timestamp.
phase: Short workstream or phase name.
decision: The choice made.
why: The constraint or tradeoff that led to it.
evidence: A compact pointer such as a file and line, command result, issue, check run, screenshot, or artifact path.
result: What happened after the choice.

Location and scope

Default to an untracked scratch path for operational notes. Put the ledger in the repository only when the user asked for a durable decision trail or the repository's review contract requires it. Follow existing evidence-directory conventions when they exist.

Log only decisions from the active task that you directly observed. Do not add routine commands, commentary updates, or reconstructed reasoning. Never place secret values, customer data, raw prompts, private messages, or full tool output in the ledger.

Review

Before handing back:

Compare the ledger with the current task's explicit plan, changed files, command results, and proof artifacts.
Remove any row whose evidence does not support the stated decision or result.
Add a missing consequential decision only when the current task still contains direct evidence for it.
Confirm every evidence pointer exists or is a stable external identifier.
State whether the ledger is untracked, committed, or intentionally omitted.

The final report should summarize the outcome. The ledger is supporting proof, not a substitute for judgment.

---name: show-me-your-workdescription: Keep a compact, reviewable decision ledger for long-running or unattended work. Use when a reviewer needs to reconstruct consequential choices and their evidence without reading private transcripts or replaying the entire task.---# Show me your workKeep one decision ledger for the current task when a human will review the workafter the fact. Record choices that changed scope, design, safety, verification,or disposition. Do not turn the ledger into a second transcript.This workflow adapts pstack's `show-me-your-work` without transcript mining.Never read unrelated chats, raw session logs, browser history, or privaterecords to reconstruct the ledger.## Ledger formatUse a TSV file with this header:```textts\tphase\tdecision\twhy\tevidence\tresult```Each cell stays on one line. Prefix spreadsheet-formula starters (`=`, `+`,`-`, or `@`) with a single quote when content is not fully controlled.- `ts`: ISO 8601 timestamp.- `phase`: Short workstream or phase name.- `decision`: The choice made.- `why`: The constraint or tradeoff that led to it.- `evidence`: A compact pointer such as a file and line, command result, issue, check run, screenshot, or artifact path.- `result`: What happened after the choice.## Location and scopeDefault to an untracked scratch path for operational notes. Put the ledger inthe repository only when the user asked for a durable decision trail or therepository's review contract requires it. Follow existing evidence-directoryconventions when they exist.Log only decisions from the active task that you directly observed. Do not addroutine commands, commentary updates, or reconstructed reasoning. Never placesecret values, customer data, raw prompts, private messages, or full tool outputin the ledger.## ReviewBefore handing back:1. Compare the ledger with the current task's explicit plan, changed files, command results, and proof artifacts.2. Remove any row whose evidence does not support the stated decision or result.3. Add a missing consequential decision only when the current task still contains direct evidence for it.4. Confirm every evidence pointer exists or is a stable external identifier.5. State whether the ledger is untracked, committed, or intentionally omitted.The final report should summarize the outcome. The ledger is supporting proof,not a substitute for judgment.

Frequently asked questions