Project Artefacts
Generate handover documents that capture intent and decisions. These artefacts are read by the Coding Agent and persist beyond chat history.
Why artefacts matter
They create durable decisions that persist beyond chat windows, enabling multiple sessions to work from the same source of truth. This prevents “context loss” when AI chat history gets too long or resets.
Core Docs vs Briefs
Core docs define the project — stable, foundational, rarely change.
Briefs define individual work items — created as needed, one per feature or phase.
The first brief (building the POC) is effectively the first PRD. Each subsequent brief scopes a specific piece of work.
Core Project Documents
Vision
projectai-handover-1-vision.md → project-vision.md
## Handover Prompt — Project Vision Document _This prompt helps AI write a project vision document._ @AI: You are now producing the **Project Vision Document** for this project. This document is the long-lived anchor for the work that follows. It should clearly articulate **why this project exists, what success looks like, and what matters most**, while also making the intended scope of the current phase explicit. **Stability note:** This prompt is intentionally stable. Only update it when the required structure, voice rules, or recurring AI failure modes indicate the template itself needs correcting. This is not a requirements document. This is not an implementation plan. This is a **decision-guiding artefact**. --- ## Purpose of the Vision Document The Vision Document should: - Align all stakeholders on the project’s true intent - Provide a shared definition of success - Guide trade-offs and decision-making during delivery - Prevent feature-led or solution-led drift - Clarify how a focused POC or MVP fits within a broader ambition A reader should be able to understand: > “Why are we doing this, who is it for, what are we trying to prove, and what are we *not* trying to do yet?” --- ## Required Structure ### 1. Project Purpose & Context - What this project is fundamentally about - The problem or opportunity it exists to address - The broader context (organisational, cultural, market, or operational) in which it sits Focus on *intent*, not solutions. --- ### 2. North Star & Desired Outcomes - The core direction this project is aiming toward - The change this project is intended to create - What meaningful success looks like if this project goes well This should describe outcomes, not outputs. --- ### 3. Primary Personas & Core Needs - Who this project is primarily for - Key personas or audiences involved - The core needs, jobs-to-be-done, or tensions for each Avoid edge cases. Focus on the users that matter most for this phase. --- ### 4. Decision Principles List the principles that should guide decisions when trade-offs arise. Examples: - What should be prioritised over speed, polish, or completeness - What values, constraints, or risks must be respected - What matters most when things compete (e.g. clarity vs flexibility) These principles should be usable during delivery. --- ### 5. Prototype / Phase Success Scenarios Define **1–3 concrete success scenarios** for the current phase (POC or MVP). For each scenario, include: - Persona - Trigger or starting point - High-level steps - What success looks like when viewed or demonstrated These scenarios should represent the **minimum convincing proof** that the project is on the right path. --- ### 6. Vision vs Current Phase Scope Explicitly separate: - **The broader vision** (where this could go over time) - **The scope of the current phase** (what this work is responsible for proving or delivering) Be clear about: - what is intentionally included now, - what is intentionally excluded or deferred, - and why that focus is appropriate at this stage. This section is critical for alignment. --- ### 7. Non-Goals & Boundaries Clearly state: - What this project is *not* trying to solve - What success does *not* require at this stage - Any known constraints or boundaries that should not be crossed This protects the vision by limiting dilution. --- ## Style & Tone Requirements - Clear, grounded, and confident - Write in **UK English suitable for New Zealand** (unless instructed otherwise) - Plain English, direct and friendly - No marketing language or hype - No technical implementation detail - Written as a trusted partner aligning a team --- ## Final Check Before Output Before responding, ensure that: - The vision is clear without prescribing solutions - The current phase scope is explicitly defined - A delivery or coding agent could use this to guide decisions - A stakeholder could read this and understand *why this project matters* Now produce the **Project Vision Document** in markdown code.
Brief Requirements (PRD)
Use this prompt to create individual briefs for features or phases of work. Briefs are saved to /catalyst/briefs/ with state prefixes.
Brief (PRD)
brief-requirements.md → catalyst/briefs/{state}-{date}_{name}.md
## Handover Prompt — Brief Requirements (PRD)
_This prompt helps AI write a Brief — a Product Requirements Document for a specific feature or phase of work._
@AI: You are now producing a **Brief** (Project Brief / PRD).
**Stability note:** This prompt is intentionally stable. Only update it when the required structure, voice rules, or recurring AI failure modes indicate the template itself needs correcting.
---
## Understanding Briefs vs Core Docs
**Core Project Documents** (in `/catalyst/specs/`):
- `project-vision.md` — North star, success criteria, decision principles (stable, rarely changes)
- `project-experience.md` — Users, journeys, features (occasionally updated)
- `project-brand.md` — Voice, visuals, communication style (stable)
- `project-architecture.md` — Technical stack, patterns, conventions (ongoing)
**Briefs** (in `/catalyst/briefs/`):
- Individual PRDs for specific features, phases, or bodies of work
- Created as needed throughout the project
- The first brief (building the POC) is effectively the first PRD
- Each brief defines what will be delivered, with boundaries and acceptance criteria
**Key distinction:** Core docs define the project. Briefs define individual work items.
---
## Purpose of a Brief
A Brief (PRD) should:
- Translate the project vision and architecture into an actionable scope for a specific feature or phase
- Clearly define what "done" means for this work
- Prevent scope creep and ambiguity during build
- Enable confident estimation, sequencing, and review
A reader should be able to understand:
> "What are we building in this brief, for whom, within what constraints, and how will we validate it?"
---
## Step 1: Declare the Phase Type (If Applicable)
If this brief represents a major phase (POC, MVP, MMP, Pilot, Production), state it at the top:
- **POC** (proof-of-concept: prove feasibility / direction)
- **MVP** (minimum viable product: smallest valuable usable release)
- **MMP** (minimum marketable product: ready for external release / adoption)
- **Pilot** (controlled rollout to validate value and operations)
- **Production v1** (hardened, supported release)
For feature briefs within a phase, reference which phase you're working in and what this feature contributes.
Then define:
- The primary purpose of this brief
- What this work is intended to prove or deliver
- What would indicate successful completion
---
## Required Structure
### 1. Summary
- One-paragraph summary of what will be delivered in this brief
- Primary users / stakeholders
- The core outcome this work is designed to achieve
---
### 2. Objectives & Success Criteria
List clear objectives for this work.
For each objective, include success criteria that are:
- observable
- testable at prototype level
- appropriate to the phase type
Examples of success criteria:
- "A stakeholder can complete the primary journey end-to-end in under 2 minutes."
- "The POC demonstrates the key workflow without needing explanation."
- "The pilot captures required feedback signals."
---
### 3. Users, Roles & Permissions (Phase-Specific)
List the user roles included in this work.
For each role:
- Key goals for this phase
- Critical tasks they must be able to do
- What is explicitly NOT required for this role yet
If authentication/permissions are not in scope, say so clearly and describe how role-based testing will occur (e.g. demo switcher, mocked profiles).
---
### 4. Scope Definition (Hard Boundaries)
#### In Scope (Must Deliver)
Provide an explicit list of what will be delivered, grouped logically:
- experiences / journeys
- modules / sections
- screens / pages (where relevant)
- key interactions
Use clear language and avoid implementation detail.
#### Out of Scope (Must NOT Deliver)
List what is explicitly excluded from this work, including:
- features
- integrations
- data persistence
- authentication and security hardening
- automations
- reporting/analytics
- edge cases
Also include "Not now" items that are tempting but intentionally deferred.
---
### 5. Requirements (Execution-Ready)
Write requirements as clear statements, grouped by area (not by speculation).
For each area:
- Provide required behaviours
- Provide required states (view/edit/error/empty if relevant)
- Provide any data assumptions (mocked, seeded, static, placeholder)
Keep requirements:
- specific enough to implement
- not so detailed that they prescribe technical solutions
Where appropriate, use a format like:
- **R#** Requirement statement
- **Notes:** clarifying context or boundaries
---
### 6. Non-Functional Requirements (Appropriate to Phase)
List constraints that apply to this work, such as:
- performance expectations (lightweight, demo-ready, etc)
- accessibility expectations (baseline vs production)
- responsiveness (mobile/desktop)
- data/privacy constraints
- hosting/deployment expectations (if discussed)
- brand/tone expectations (if applicable)
Be explicit about what is "prototype acceptable" vs "production required later".
---
### 7. Design & Content Requirements
Include what is required for:
- design direction (principles, tone)
- content approach (real content vs placeholder)
- trust cues (where users need clarity/confidence)
- any sensitive considerations (culture, governance, identity, language)
Do not turn this into a design system — focus on what matters for this work.
---
### 8. Risks, Assumptions & Open Questions
Provide:
- Key risks that could derail this work
- Assumptions the scope depends on
- Open questions that must be resolved
Mark each open question as one of:
- **Blocker** (must be answered before build)
- **During Build** (can be decided while implementing)
- **Post-Phase** (can be deferred safely)
---
### 9. Acceptance Checklist
Provide a checklist that can be used to validate completion.
Must include:
- required journeys demonstrable end-to-end
- required screens/sections exist and are navigable
- scope boundaries respected (no "extra features")
- success scenarios from the Vision are demonstrable
- known exclusions remain excluded
- clear handoff notes for next phase
For POC/MVP, prioritise demonstrability over completeness.
---
### 10. Handoff Notes (For Implementation Agent)
Add a short section that helps an implementation-focused agent:
- what to prioritise first
- what to be careful not to overbuild
- what assumptions to confirm early
- what "good" looks like for a first iteration
Do not include code or repo-specific instructions.
---
## Brief Callout (Required)
Every brief **must** include this callout immediately after the title:
```markdown
> **Catalyst Brief** — Read [AGENTS.md](../../AGENTS.md) and [BRIEFS.md](../BRIEFS.md) before implementing.
```
This ensures AI agents know to check the workflow documentation before starting work.
---
## Brief Naming Convention
Save the resulting brief to `/catalyst/briefs/` using the naming convention:
```
{state}-{date}_{brief-name}.md
```
**States:** `backlog` (not started), `approved` (ready to start), `active` (in progress), `_review` (done, needs review), `_blocked` (stuck)
**Examples:**
- `backlog-20260107_user-authentication.md`
- `approved-20260107_payment-integration.md`
- `active-20260107_dashboard-redesign.md`
See `AGENTS.md` for the full brief workflow.
---
## Style & Tone Requirements
- Clear, structured, and unambiguous
- Write in **UK English suitable for New Zealand** (unless instructed otherwise)
- No fluff, no hype
- Use headings and bullet points
- Prefer explicit boundaries over implied intent
---
## Final Check Before Output
Ensure:
- Phase type is clearly stated (if applicable) and reflected in the level of detail
- In-scope and out-of-scope are explicit and strict
- Requirements are implementable without guessing
- Open questions are clearly labelled by urgency
- Another AI could implement this without re-reading the full chat history
Now produce the **Brief** in markdown format.
Document hierarchy
Core docs define the project — created once, updated rarely.
Briefs (PRDs) are created as needed — one per feature or phase.
Where artefacts live
The Coding Agent reads core docs when starting a session. Briefs provide specific scope for individual work items.
When to create each
Vision
At project start, after initial alignment conversations
Architecture
After key technical decisions are made
Voice
When tone/messaging consistency matters (optional)
Brief (PRD)
Before each piece of work — features, phases, or tasks
Quality tips
- • Keep docs focused and concise — AI context windows are limited
- • Update core docs when decisions change — don't let them drift
- • Briefs should be scoped tightly — one brief per feature or phase
- • Core docs are long-lived; briefs are task-lived
Common mistakes
- • Creating docs that are too long — bloats AI context
- • Not updating core docs when decisions change
- • Writing briefs before vision is clear
- • Skipping architecture — leads to inconsistent Coding Agent decisions
- • Treating briefs as “requirements in specs/” — they belong in briefs/
Related docs
- Project Sessions — Start and continue project conversations
- Coding Sessions — The Coding Agent reads these artefacts
- Delivery Cycles — Artefacts are reviewed at Brief checkpoint