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/design-oq-triage.md
View on GitHub---
name: Design OQ Triage Protocol
description: Two-tier classification of open questions from Research — viability OQs answered before Evaluate matrix is built; elaboration-input OQs deferred
type: reference
---
# Design OQ Triage Protocol
Open questions (OQs) surfaced during Research fall into two tiers that determine when they must be answered.
## Two Tiers
| Tier | Label | Definition | When to resolve |
| -------------------- | ----- | ------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------- |
| **Viability** | `(V)` | Would the answer disqualify one or more options from the matrix? (e.g. legal gate, hard network constraint, missing required feature) | **Before** the Evaluate matrix is built |
| **Evaluation-input** | `(E)` | Affects cost modelling, sizing, or implementation detail — does not change which options are viable | Defer to Evaluate Step 3 or capture as implementation tickets |
## Interaction format
Propose a classification, not a question. For each OQ, state the agent's own
best `(V)`/`(E)` call with a one-line reason, and present every OQ together —
this is a draft to confirm, not a Q&A session:
1. **Classify every OQ up front**, reasoning from the option set and matrix
dimensions already in hand — most OQs' tier is evident without asking.
2. **Present the full list in one message**, at the phase's sign-off
(`patterns/phase-sign-off.md`), alongside everything else the developer is
confirming that phase — not as a separate stop.
3. **The developer corrects what's wrong** in their one reply; anything they
don't comment on is accepted as proposed.
Do not turn this into a separate one-at-a-time interactive round — that was
the old model, and it is exactly the per-step stop the lifecycle's gate
reduction removed. If a specific OQ's classification is genuinely unclear
even after reasoning about it, flag that one by name in the sign-off message
rather than guessing silently — never invent a (V)/(E) call with no stated
basis.
## Rules
1. After Research Step 2 (options returned), propose a `(V)`/`(E)` call for
every OQ (Step 3b) and present it at Step 5's sign-off alongside the
option set, per the interaction format above.
2. Every `(V)` OQ needs a resolution recorded (or the affected option marked
conditional) before `design/01 Research.md`'s sign-off is approved.
3. At the start of Evaluate (Step 1b), re-read the OQ list. If any unresolved
`(V)` OQs remain, this is an intelligent-approval checkpoint
(`patterns/approval-intelligent.md`) — auto-continue when none remain,
escalate by name when they do.
4. In Solution, re-check whether any `(V)` OQs from Evaluate are still open.
If any remain unresolved and would block implementation, create a spike
ticket for them in `Efforts.md`.
5. Capture `(E)` OQs in `design/01 Research.md`'s Open Questions table with
`(E)` prefix — they feed Evaluate Step 3's recommendation, which lists
open questions with a suggested owner for each.
## Why
When many OQs surface during Research and the recommendation is written
before viability-blocking OQs are resolved, the recommendation must be
revised mid-Evaluate once the blockers surface. Tiering OQs at Research —
enforcing that `(V)` OQs are resolved before the matrix is built — prevents
this rework. Proposing the classification instead of asking through it,
folded into the phase sign-off, is what keeps this a confirmation instead of
its own approval gate.
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: 472k | 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 |