| name | impact-map |
| description | Capture a user story, actor, action, goal, and business rules from the user. Use when the user wants to start a new feature and has not yet provided structured business context, or types /impact-map. |
| phase | 1-input |
| parallel | false |
| inputs | [] |
| outputs | ["context-md"] |
| escalation | none |
/impact-map
Elicit and structure the business context behind a user story before any code or scenarios are written. This is the "board d'impact mapping" of the pipeline: actor → goal → action → user story → business rules.
When to use
- Start of a new feature.
- The user pastes a half-formed story like "we need cart checkout" with no rules.
- Upstream of
spec-generate.
Inputs
None required. Pull everything from the user via grilling.
Workflow
1. Grill for the impact map (mandatory)
Ask the user, in order:
- Actor. Who triggers this? Be specific (role, not "user").
- Goal (impact). What outcome does the actor want? Phrase as a business impact, not a UI behavior.
- Action. What does the actor do to reach the goal?
- User story. Compose:
As a <actor>, I want to <action>, so that <goal>.
- Business rules. Enumerate every rule that constrains the action. Press for:
- Authorization rules (who can/cannot).
- Boundary / limit rules (max amount, time windows).
- Conflict rules (what's incompatible).
- Lifecycle / state rules (pre/post conditions).
- Target milestone / version (optional).
- Non-goals. What is explicitly out of scope?
Stop and ask if any answer is vague. Do not infer rules from your training; surface ambiguities as questions.
2. Emit context.md
Write specs/<us-slug>/context.md with this structure:
# <us-slug>
## Actor
<role>
## Goal
<business outcome>
## Action
<what the actor does>
## User story
As a <actor>, I want to <action>, so that <goal>.
## Business rules
- <id: short label> — <rule>
- ...
## Non-goals
- ...
## Milestone
<version or "TBD">
Rule IDs follow R-<NN> (e.g. R-01). They will be reused as tags on Gherkin scenarios.
Output
specs/<us-slug>/context.md
Handoff
Next skill in the pipeline: spec-generate. It reads context.md and produces Gherkin scenarios per rule.
Composition with mattpocock skills
If the project already uses the mattpocock skill set:
grill-with-docs is a stronger interview engine than the plain grilling described above; invoke it as a sub-tool to gather rules and update CONTEXT.md / ADRs in the same session, then transcribe the result into specs/<us-slug>/context.md.to-prd produces a sibling PRD document. If used, link the PRD path from context.md.
