# Give an engineering investigation an exit decision

Illustrative planning brief; no automatic product import.

Illustrative uncertainty: the team does not know whether the existing data access can support a scoped report filter.

## Decision

Collect enough evidence to choose reuse, a bounded change or a revised slice before scheduling implementation.

## Owned work

- Name the uncertainty
  - Owner role: Technical lead
  - Acceptance evidence: The investigation question affects the planned slice.
- Collect bounded evidence
  - Owner role: Engineer
  - Acceptance evidence: A safe test or source inspection answers the relevant feasibility question.
- Record the exit decision
  - Owner role: Planner
  - Acceptance evidence: The outcome chooses a next route and identifies remaining unknowns.

## Workflow

1. State the planning decision blocked by uncertainty.
2. Choose the smallest investigation that could change it.
3. Close with evidence and a route choice, not a large unrelated prototype.

## Judgment

A prototype or local test does not establish production availability or all operational conditions.


## Illustrative delivery brief

Uncertainty: support for the agreed record boundary.
Evidence needed: a bounded source/test observation.
Exit choices: reuse, scoped change or revise the slice.
Not committed: a complete report redesign.


## Workflow questions

### What if the spike cannot answer everything?

Record the remaining unknowns and decide whether they block the next bounded action.

### Does a spike always produce code?

No. Source inspection or a short evidence memo may be sufficient to change the planning decision.

## Product connection

Use the investigation’s decision and exit evidence to review a TeamBoostAI task proposal before committing implementation work.

Confirm account availability before adopting this manual outline.