Bespoke prompt architect

Purpose & trigger

Turn an objective into a precise, bounded, verifiable agent mission with authority, constraints, checkpoints, and evidence. Use when writing implementation prompts, technical missions, runbooks, acceptance criteria, or agent handoff instructions for consequential work.

Source

Portable instructions served exactly from the validated skill package.

---
name: bespoke-prompt-architect
description: Turn an objective into a precise, bounded, verifiable agent mission with authority, constraints, checkpoints, and evidence. Use when writing implementation prompts, technical missions, runbooks, acceptance criteria, or agent handoff instructions for consequential work.
---

# Bespoke Prompt Architect

## Specify the mission

1. State the outcome, governing authority, repository or system boundary, and audience.
2. Separate requirements, non-goals, immutable constraints, and assumptions that need verification.
3. Name the allowed and prohibited writes, external actions, irreversible decisions, and escalation conditions.
4. Break execution into reviewable checkpoints with concrete acceptance criteria, exact verification commands, and honest handoff fields.
5. Remove duplicate capability lists, vague completion language, untestable claims, and instructions that conflict with the governing specification.

## Make the prompt auditable

- Require evidence before claims and fresh verification before completion.
- Prefer observable contracts over desired implementation details where multiple safe solutions exist.
- Do not grant deployment, publication, DNS, secrets, or destructive authority implicitly.

Bundled files

• SKILL.md — Portable prompt-authoring instructions.

Verification evidence

• Structural validation and representative exercise record.

Known limitations

• Prompt quality still requires task-specific human review.