nhouseholder/type-aware-rigor

Type-Aware Rigor — Match Depth to Work Type

When This Fires

During brainstorming, planning, QA, or deployment decisions. This skill overrides default "size-based" heuristics with "type-based" ones across multiple downstream skills.

Rigor Levels

| Project Type | Rigor | Brainstorm | QA Tier | Plan Detail | Deploy Gate | |-------------|-------|------------|---------|-------------|-------------| | Site update | Tight | Lite only | Tier 1 (spot-check) | Bullet list | Visual verify | | New feature | Standard | Lite or Full | Tier 2 (functional) | Step-by-step | Baseline + verify | | New build | Deep | Full mandatory | Tier 3 (comprehensive) | Detailed with alternatives | Full smoke test | | Algorithm/model | Deep | Full mandatory | Tier 3 + data validation | Detailed + hypothesis | Backtest + verify | | Campaign/one-off | Creative | Lite | Tier 1 (it works) | Minimal | Run once, confirm |

How to Detect Type

Look at the user's request + project context:

• "Add dark mode" in an existing app → Site update (Tight)
• "Build a new prediction model" → Algorithm (Deep)
• "Create a new site for X" → New build (Deep)
• "Write a script to migrate the data" → Campaign (Creative)
• "Add a leaderboard feature" → New feature (Standard)

When ambiguous, default to Standard.

Downstream Influence

Brainstorming

• Tight/Creative → force Lite brainstorm regardless of file count
• Deep → force Full brainstorm regardless of apparent simplicity
• Standard → use brainstorming's existing complexity gate

QA Gate (qa-gate skill)

• Tier 1 (Tight/Creative): Does it work? Quick spot-check. No regression sweep.
• Tier 2 (Standard): Functional test of the new feature + baseline verification of surrounding features. Check one happy path and one edge case.
• Tier 3 (Deep): Full regression across all affected pages/components. Data validation. Visual comparison. Load the actual page and verify every element.

Planning Depth (writing-plans skill)

• Tight: Bullet list of changes. No alternatives section. Skip dependency analysis.
• Standard: Step-by-step with file paths and commands. Brief consideration of alternatives.
• Deep: Detailed plan with alternatives analysis, risk assessment, rollback strategy, and explicit verification criteria for each step.
• Creative: Minimal plan — just enough to not lose track. Speed > thoroughness.

Deploy Decisions (deploy skill)

• Tight: Deploy after visual verification of changed pages only.
• Standard: Deploy after baseline comparison of changed pages + spot-check of related pages.
• Deep: Full smoke test of all pages. Data consistency check. Version regression check. Canary period if available.
• Creative: Deploy and verify once. No canary needed for throwaway code.

The Key Insight

The COST of getting it wrong scales with project type:

| Type | Cost of a Bug | Cost of Over-Planning | |------|--------------|----------------------| | Site update | Low (quick fix) | Medium (wasted time) | | New feature | Medium (users affected) | Low (better safe) | | New build | High (architecture debt) | Low (investment pays off) | | Algorithm | Very high (weeks of bad data) | Very low (always worth it) | | Campaign | Very low (throwaway code) | High (defeats the purpose) |

Match your effort to the cost of failure, not the size of the change.