Lists multiple specific concrete actions: 'Creates structured specifications', 'validates execution plans against specs', 'manages two-loop planning-to-execution workflows with decision gates', and 'captures provenance of AI-generated work'.

Clearly answers both what (creates specs, validates plans, manages workflows, captures provenance) AND when with explicit 'Use when...' clause covering multiple trigger scenarios including problem-based triggers ('agent keeps building the wrong thing') and keyword triggers.

Excellent coverage of natural terms users would say: 'spec-driven', 'SDD', 'specification first', 'code is a side effect', 'agent keeps building the wrong thing', 'decision gates', 'provenance', 'AI coding agents', 'non-trivial feature'.

Clear niche focused on spec-driven development with AI agents, two-loop workflows, and decision gates. The specific terminology ('SDD', 'spec-driven', 'provenance', 'two-loop') creates distinct triggers unlikely to conflict with general coding or documentation skills.

The content is well-written but includes some philosophical framing ('Code is a generated side effect', 'Cognitive budget replaces developer-hours') that, while valuable for understanding, adds tokens beyond pure instruction. The core workflow could be tighter.

Provides concrete, numbered steps for both loops, explicit spec format with required sections, clear decision gate criteria, and specific guidance like 'three prompts' for execution. The spec template sections are copy-paste ready.

Excellent workflow structure with nine clearly numbered steps across two loops, an explicit decision gate at step 6 with clear branching logic (fix spec vs fix plan), and mandatory validation checkpoints. The 'When NOT to Use SDD' section adds appropriate boundaries.

Well-structured with clear overview sections and appropriate references to external files (references/principles.md, references/anti-patterns.md) for detailed content. The main document stays focused on the workflow while pointing to deeper material.