Twelve-Factor Agentic SDLC — Prompts
Excerpt 1 — Factor I: Strategic Mindset Checklist
From content/strategic-mindset.md:
# I. Strategic Mindset: Developer as Orchestrator, AI as Intern
*The foundation of the Agentic SDLC is a strategic mindset shift. Treat the AI as a fast,
knowledgeable junior partner that requires clear direction, mentorship, and rigorous review.
Your role is elevated from writing code to orchestrating solutions.*
- [ ] **Adopt the Mantra:** Your primary focus becomes "Debug the spec, not the code."
If the AI generates incorrect code, the first step is to refine the specification or
the context provided, not just to fix the output.
- [ ] **Mentor the Intern:** When the AI makes a mistake, correct it by providing clear
feedback and better examples, similar to how you would guide a junior developer.
- [ ] **Own the Outcome:** You are the orchestrator. The AI is a tool, and you are ultimately
responsible for the quality, security, and architectural cohesion of the final product.
Prompting technique: Not a prompt to an AI — this is a checklist for a human developer's mental model. The "Developer as Orchestrator, AI as Intern" frame is a reframing technique for human cognition, not a system prompt. The [ ] checkbox format makes it actionable rather than abstract.
Excerpt 2 — Manifesto Failure Mode Analysis
From manifesto.md:
The Agentic SDLC was born from a critical observation: while individual developers are
experiencing personal productivity gains with AI tools, engineering teams are failing to
translate these wins into a collective increase in velocity. Ad-hoc prompting and
inconsistent workflows, or "Vibe Coding," often lead to systemic problems.
**Inconsistent Team Output:** Different prompting styles and a lack of shared standards
lead to a chaotic codebase and unpredictable quality. One developer's shortcut becomes
the entire team's technical debt.
**Organizational Knowledge Gap:** Critical knowledge about prompts, patterns, and
architectural standards remains siloed with individuals. The team's collective intelligence
never improves, leading to duplicated work and repeated mistakes.
Prompting technique: Framing-by-negation — the methodology is defined through what goes wrong without it. This is a persuasion pattern (problem → solution) applied to methodology documentation.
Excerpt 3 — Factor III: Mission Definition Procedure
From content/mission-definition.md:
- [ ] **Start with the Mission Brief:** The process begins when you are assigned a task
in the issue tracker that contains a Mission Brief (Goal, Constraints, Success Criteria).
- [ ] **Generate the Specification (spec.md):** Your first action is to use the platform's
CLI (e.g., /specify) to generate a detailed, version-controlled spec.md file
from the Mission Brief.
- [ ] **Link the Artifacts:** Commit the newly created spec.md to the feature branch and
post a link to it back in the issue tracker, ensuring traceability from intent
to specification.
Prompting technique: Prescriptive checklist with numbered ordering and conditional inputs. The "Mission Brief" → spec.md artifact chain is presented as a manufacturing process — each step produces a versioned artifact that feeds the next.