APS Terminology
This page defines the words APS uses and what they mean.
Renamed concepts (v0.2.0+)
| Old term | Current term | Meaning |
|---|---|---|
| Task | Work Item | A bounded unit of work with intent, outcome, scope, and validation |
| Step (file) | Action Plan | Execution breakdown for a work item |
| Step (numbered) | Action | A coherent unit of execution within a plan |
| Checkpoint | Checkpoint | Observable proof that an action is complete |
New projects start with current terminology. Upgrading from pre-v0.2 projects:
- Task sections → Work Items
.steps.mdfiles →.actions.md- Numbered steps → Actions
No content needs rewriting unless it violates checkpoint rules.
Core principle
- Specs describe intent.
- Work items define outcomes.
- Action plans decompose execution.
- Checkpoints verify state.
Implementation emerges from patterns and agent judgement.
Hierarchy
| Layer | Purpose | You write | You do not write |
|---|---|---|---|
| Index | Plan overview | Modules, milestones, risks | Implementation detail |
| Module | Bounded work area | Interfaces, boundaries, work items | Code snippets |
| Work Item | Outcome unit | Intent, outcome, validation | How to implement |
| Action Plan | Execution breakdown | Actions, checkpoints | Tutorials |
| Action | Unit of work | Purpose, produces, checkpoint | Code structure |
| Checkpoint | Verification | Observable state | Implementation steps |
Work items
A work item is the contract: what will be achieved and how success is verified.
Required fields
- Intent — One sentence describing the outcome
- Expected Outcome — Testable or observable result
- Validation — Command or condition to verify completion
Optional fields
- Scope / Non-scope
- Dependencies
- Confidence (
low,medium,high) - Files (best-effort)
- Risks
- Deliverables
Rules
- Work items describe what must be true, not how it is achieved
- Work items authorise change but do not prescribe implementation
- Validation must be deterministic where possible
Action plans
An action plan decomposes a work item into executable actions.
When to create one
- The work item is non-trivial
- Multiple artefacts are produced
- Ordering or dependencies matter
Simple work items may be executed without an action plan.
File naming
plans/execution/{WORK-ITEM-ID}.actions.md
Example: AUTH-001.actions.md
Actions and checkpoints
Each action represents a coherent unit of execution, not a tutorial.
## Action 1 — Create authentication middleware
**Purpose** Validate requests and attach user context.
**Produces** Auth middleware module.
**Checkpoint** Auth middleware validates requests, attaches user to context
**Validate** `npm test -- auth.middleware.test.ts`
Checkpoint rules
- Must be verifiable by inspection or command
- Must avoid implementation detail
- Must remain valid even if implementation changes
✅ Good: "Authenticated requests attach user context"
❌ Bad: "Create mapping.ts with switch statement"
Module types
Most modules are vertical — they own a single domain end-to-end.
A conductor (crosscutting) module coordinates work across vertical modules
— release cuts, security audits, perf budgets. Mark it with Type: Conductor in
the metadata table. See
Conductor modules
in the source repository.
Next: Taxonomy →