when-using-sparc-methodology-use-sparc-workflow

STANDARD OPERATING PROCEDURE

Purpose

Run SPARC methodology end-to-end with clear intent capture, stage gates, evidence-backed reviews, and confidence-aware delivery.

Trigger Conditions

• Positive: structured problem solving using SPARC, stage-gated delivery, retrospectives, consolidation of learnings.
• Negative: ad-hoc single-pass tasks, prompt-only edits (route to prompt-architect), or new skill weaving (route to skill-forge).

Guardrails

• Skill-Forge structure-first: maintain SKILL.md, examples/, tests/; add resources//references/ or log remediation tasks.
• Prompt-Architect hygiene: capture HARD/SOFT/INFERRED constraints per SPARC stage, keep English-only outputs, and declare ceilings.
• Stage safety: set entry/exit criteria for Scope/Plan/Act/Review/Consolidate, enforce registry usage, and keep hook latency budgets.
• Adversarial validation: challenge assumptions each stage, run COV, and document evidence and deltas.
• MCP tagging: store SPARC runs with WHO=sparc-workflow-{session} and WHY=skill-execution.

Execution Playbook

1. Scope: define objective, constraints, and success metrics; confirm inferred assumptions.
2. Plan: design approach, assign owners, and set timelines plus rollback points.
3. Act: execute tasks with monitoring, TodoWrite updates, and guardrails.
4. Review: validate outcomes, run adversarial checks, and log evidence.
5. Consolidate: capture learnings, decisions, and next actions.
6. Delivery: summarize SPARC path, evidence, risks, and confidence ceiling.

Output Format

• SPARC stage summary with constraints and decisions.
• Evidence log, risks, and follow-ups.
• Confidence: X.XX (ceiling: TYPE Y.YY) - rationale.

Validation Checklist

• Structure-first assets present or ticketed; examples/tests reflect SPARC flow.
• Stage gates and rollback points defined; registry and hooks validated.
• Adversarial/COV runs stored with MCP tags; confidence ceiling declared; English-only output.

Completion Definition

Workflow is done when SPARC stages complete with evidence, risks are owned, learnings captured, and MCP logs tagged for reuse.

Confidence: 0.70 (ceiling: inference 0.70) - SPARC workflow doc rewritten with skill-forge scaffolding and prompt-architect constraint/confidence discipline.