# Software-factory feedback review

Working template. Fill this with actual source links before making a process change.

## Decision

- Repository or workstream:
- Recurring problem:
- Decision owner:
- Proposed action:
- Actions considered and why not chosen (smaller PRs, reviewer assignment, training, rule):
- Why this action fits the evidence:

## Evidence and scope

- Baseline start/end dates and timezone:
- Team, repository and work types included:
- Tool, attribution and cost-source coverage:
- Relevant PRs/reviews/deployments:
- Metric name, unit, denominator, endpoint, timing mode and weekend treatment:
- Baseline value(s) per metric:
- Investigation questions (for example, Wooly) and source links followed:
- PRs and review comments showing the recurring problem:
- Known missing data or immature follow-up windows:
- Other explanations for the observed pattern:

## Repository standard

- Standard filename:
- Rule commit or revision:
- Checks version, intelligence setting, files glob (if any), run mode (Actions or local), fail-on-findings value, model or Router use:
- Intended behavior:
- Concrete problem to flag:
- Evidence required for a finding:
- Legitimate exceptions:
- Exception owner and evidence required for an exception:
- Repository context needed:
- Engineer responsible for reviewing the rule:

## Rule validation

| Actual example and run link/artifact | Expected flag or allow | Observed flag, allow, skipped, error, unusable result or not run | Finding was useful | Reason and correction, including skip/error cause |
| --- | --- | --- | --- | --- |
| | | | | |
| | | | | |
| | | | | |

- Representative run cost and cost source:
- Applicable rule revision for each example (if different from the revision above):
- Rejected or noisy findings:
- Changes to the standard:
- Existing tests/CI that exercise the behavior:
- Separate merge/branch-policy decision and owner:

## Follow-up

- Follow-up start/end dates and timezone:
- Scope or work-mix differences from baseline:
- Follow-up value(s) per metric, using the same definitions as baseline:
- Date the 90-day turnover window closes for the last included PR:
- Deployment matching coverage for both periods:
- Did the same concern recur in later PRs?
- What changed in the original metric and underlying work?
- Did the rule introduce noise or extra work?
- Which quality follow-up windows are still incomplete?
- Continue, revise or remove the rule:
- Reason and next review date:

Keep code-effort estimates, recorded time, paid cost and modeled return separate. A before-and-after association alone does not prove the rule caused the change.
