fpf-decision-record

Output

.fpf/decisions/DRR-${CLAUDE_SESSION_ID}--<slug>.md

Constraints (quality bar)

• C1: Bounded context stated
• C2: ≥2 real alternatives enumerated (including "do nothing")
• C3: Each option has costs, risks, second-order effects
• C4: Decision states what was chosen AND why (trade-offs explicit)
• C5: Evidence linked or gap flagged
• C6: Rollback/migration plan if decision might be reversed
• C7: What future evidence would invalidate this decision
• C8: Production-affecting decisions MUST have deployment plan (canary strategy, success indicators, emergency stop criteria)
• C9: valid_until set — decisions go stale as context changes. Set expiry date or "perpetual" with justification.
• C10: Link to PAR-* (parity report) and SEL-* (selection record) when they exist — decisions must be traceable to comparison evidence
• C11: WLNK analysis — what is the weakest link in the chosen option? Does this change the system's current weakest link?
• C12: MONO justification — if the decision adds components, abstractions, or dependencies, state what new weak links are introduced and why the benefit justifies them

Format

# Decision Record
- **ID:** DRR-...  **Created:** YYYY-MM-DD  **Context:** ...
- **F:** ...  **G:** {contexts}  **R:** 0-1
- **valid_until:** YYYY-MM-DD
- **Parity:** PAR-...  **Selection:** SEL-...

## Options
1. [option] — costs, risks, second-order effects
2. [option] — ...

## Decision
Chosen: [option]. Rationale: ...

## WLNK analysis
- Weakest link in chosen option: ...
- Does this change the system's current weakest link? ...

## MONO justification
(If adding components/abstractions/dependencies: what new weak links, and why is the benefit worth it?)

## Evidence
...

## Rollback plan
...

## Deployment plan (if production-affecting)
- Canary strategy: ...
- Success indicators: ...
- Emergency stop criteria: ...

## Invalidation conditions
...