DataArt/artisyn-implementation-harness / skills/implementation-harness
implementation-harness
FlaggedInstall the portable AI-dev harness, rebuild its operational graph standalone, harvest project skill improvements into its canonical templates, package direct canonical edits into a portable bundle, or apply canonical updates to an existing project. Use /implementation-harness launch, /implementation-harness reindex-graph, or /implementation-harness patch harvest|prepare|apply.
canonical/skills/design/patterns/multi-decision-split.md
View on GitHub---
name: Multi-Decision Split Pattern
description: Default file-splitting policy when a design's Frame implies more than one independent decision
type: reference
---
# Multi-Decision Split Pattern
Some design tickets bundle more than one independent decision under a single epic (e.g. a security epic that bundles a triage framework, a tooling stack, and a timing decision). Treating these as a single option set produces an unwieldy EC × Options matrix and blurs which open questions and assumptions belong to which decision.
## Detecting a multi-decision Frame
Frame (`references/03-frame.md`, Step 2) checks whether the Problem/Requirements
imply two or more decisions that could each stand alone — each has its own
option set, its own tradeoffs, and could in principle be approved or reiterated
independently of the others. One decision or fewer needs no confirmation and
no `## Decisions` section. Two or more get their own dedicated sign-off, right
there in Step 2 — not folded into the main Frame sign-off (Step 3), and not
deferred, reactively, until Research is already underway. From Research
onward, every stage reads the confirmed breakdown from `design/00 Frame.md`
and applies this pattern; none of them re-detect it.
## Default policy
| Artifact | Multi-decision handling |
| ------------------------------------------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `design/01 Research.md` | One file per decision — `design/01 Research - {Decision}.md` |
| `design/02 Evaluate.md` | One file per decision — `design/02 Evaluate - {Decision}.md`, each with its own matrix, tradeoffs, and recommendation |
| `solution/Design.md` | One file per decision — `solution/Design - {Decision}.md` |
| `solution/Architecture.md` | **Single, integrated file** — the converged system view, one System Context / Component Design / Flow set per decision (only where it adds value), not one file per decision |
| `solution/Stack.md`, `solution/Efforts.md` | Single file — Efforts already spans all decisions as one WBS |
Loop the Research → Evaluate → Design content steps once per decision, using
the same skill steps each time — see `references/00-decision-router.md` for
exactly which steps loop and which close once (the stage's Sign-off → Report
→ Commit never multiply with decision count). Track each decision's OQs and
assumptions separately in `design/Report.md`'s Assumptions table (note which
decision each row belongs to).
## Why
Architecture.md is the converged deliverable — a reader needs one picture of how the decisions fit together as a system, so it stays single. Research, Evaluate, and Design are option-evaluation and per-decision-detail artifacts — bundling them blurs scope and produces an unreadable matrix. This pattern tends to get rediscovered reactively, mid-session, on the first multi-decision epic a project runs — encoding it here means it doesn't need rediscovering every time.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 470k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |