skills.rest
gmackie
gmackieCOMMUNITY
Explore all skills by this author →
Website𝕏@grahammackieGitHub

figure-it-out

Designs an auditable multi-phase playbook for large migrations and cross-cutting changes.

Updated Mar 27, 2026
Search Tags:#workflow design#migration#autonomy#audit trail#code migration#decision log#hypothesis loop
One-click install
npx skills add https://github.com/gmackie/agent-skills --skill figure-it-out-gmackie
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: figure-it-out
Source: https://github.com/gmackie/agent-skills/tree/main/skills/figure-it-out
Command: npx skills add https://github.com/gmackie/agent-skills --skill figure-it-out-gmackie

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When a task matches no existing playbook—such as a large migration across many call sites, an ambitious multi-part change, or work a human reviews after stepping away—this Skill designs a rigorous, auditable workflow before any code is written, preventing wasted effort on the wrong approach. ## Core Features & Use Cases - Workflow Design: Frames the task with a falsifiable definition of done, quantified scope, and a rigor level scaled to risk, then decomposes work into atomic, independently-landable units sequenced riskiest-first. - Hypothesis Loop: Runs each unit as an experiment—state hypothesis, make the smallest change, measure against the predicate on the real artifact, keep or revert—with verdicts of VERIFIED, NOT VERIFIED, or INCONCLUSIVE. - Audit Trail: Logs every decision and unit in a canonical TSV via the show-me-your-work skill so a reviewer can audit the run in the PR. - Use Case: You need to migrate an API across dozens of call sites. Invoke this Skill to frame the migration, build a verification harness from the pre-change baseline, execute units in parallel worktrees, and hand back a decision trail with verified results. ## Quick Start Ask the agent to figure out a plan for migrating the authentication module across all services and produce an auditable decision log.

Frequently Asked Questions about figure-it-out

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

FAQPage Schema
How do I plan a large code migration with an AI agent?▼

Invoke this Skill to frame the migration with a falsifiable definition of done, decompose it into atomic independently-landable units sequenced riskiest-first, and build a verification harness from the pre-change baseline before writing any code.

When should I use figure-it-out instead of a bug fix or feature playbook?▼

Use it when the task matches no narrower playbook, or when a normally single-unit task becomes large or cross-cutting, such as a migration across many call sites. Focused single-unit tasks should route to the Bug fix, Perf, or Feature playbooks instead.

How does the hypothesis loop verify each change?▼

Each unit states a hypothesis, makes the smallest change, and measures against the predicate on the real artifact, never a self-report. Verdicts are VERIFIED, NOT VERIFIED, or INCONCLUSIVE, and inconclusive results are not treated as passes.

What is the decision audit trail and where is it stored?▼

The audit trail is one canonical TSV logged via the show-me-your-work skill, with a row per decision and per unit and evidence as links. For ambitious work it is committed so a reviewer can read it in the PR and re-run committed evidence scripts.

Can work be parallelized across multiple agents?▼

Yes, but only across genuine seams, with each worker given its own worktree or branch. Delegated work is paired with a judge, and the orchestrator audits delegate artifacts before trusting them.

What are the limitations of this approach?▼

It is over-engineering for mechanical work whose shape is already concrete or focused single-unit tasks that fit existing playbooks. The added rigor, gates, and artifacts cost time that is only justified for high-stakes or hard-to-reverse work.

Related Skills

Explore More Software Engineering Skills→
Negentropy-LabyNegentropy-Laby
3

scope-check

Compare current implementation against original scope to flag additions and estimate effort.

Official
Intermediate
adjohnadjohn
3

principle-encode-lessons-in-structure

Converts recurring corrections into lint rules, runtime checks, and automation scripts.

Community
Basic
mmdmcymmdmcy

arena

Spawns parallel candidate solutions, judges them against a rubric, and synthesizes the strongest result.

Community
Advanced
lucasheriqueslucasheriques

poteto-mode

Routes engineering tasks to playbooks with principles, subagent delegation, and verified delivery.

Community
Advanced