process-feedback-stage2

name

process-feedback-stage2

description

Implement and operate Stage2 Process feedback capability for OperatorOne. Use when asked to build, run, audit, or deliver feedback processing from ingestion through prioritization and closed-loop tracking; generate Stage2 feedback scoreboards/alerts/weekly snapshots; create operations_to_product/marketing/sales handoffs; enforce strict reproducibility; or migrate from simulated feedback adapters to live sources.

Process Feedback Stage2

Run a deterministic Stage2 pipeline that turns raw feedback signals into prioritized actions and auditable outputs.

Execute

1. Bootstrap defaults (run once per workspace or when resetting contracts/config):
   • python3 workspaces/op1_operations/skills/process-feedback-stage2/scripts/bootstrap_stage2_feedback_contracts.py --repo-root /path/to/OperatorOne
2. Run pipeline:
   • python3 workspaces/op1_operations/skills/process-feedback-stage2/scripts/run_stage2_feedback.py --repo-root /path/to/OperatorOne
3. Confirm required outputs exist under workspaces/op1_operations/research/stage2_feedback:
   • stage2_feedback_scoreboard.latest.json
   • stage2_feedback_scoreboard.latest.md
   • weekly_feedback_snapshot.latest.md
   • reproducibility_report.latest.json
   • run_stage2.latest.json

If repo auto-detection fails, pass --repo-root explicitly.

Strict reproducibility

When reliability or delivery audit is required:

1. Set fixed timestamp:
   • export STAGE2_AS_OF="2026-03-05T12:00:00Z"
2. Run strict mode:
   • python3 workspaces/op1_operations/skills/process-feedback-stage2/scripts/run_stage2_feedback.py --repo-root /path/to/OperatorOne --strict-repro
3. Require all of:
   • reproducibility_report.latest.json.status == "passed"
   • reproducibility_report.latest.json.strict_mode == true
   • reproducibility_report.latest.json.changed_files_vs_baseline is empty

Return format

When reporting status to user, include:

1. Pipeline status (passed/failed)
2. Headline metrics (feedback_events_total, feedback_items_total, themes_total, expected_mrr_delta_30d)
3. Alert status (count + top alert codes)
4. Closed-loop status (triaged_rate, p0_untriaged, p0_unowned)
5. Reproducibility status
6. Mode note (simulated_signals vs live adapters)

Quality bar

• Keep contract alignment with contracts/feedback_contract.v1.json.
• Keep taxonomy alignment with contracts/feedback_taxonomy.v1.yaml.
• Keep weight logic alignment with config/feedback_priority_weights.v1.yaml.
• Treat missing required outputs or failed reproducibility as failed delivery.
• Treat unresolved critical alert as delivery risk.

References

• Read references/workflow.md for full run sequence.
• Read references/data-contracts.md when validating schemas/outputs.
• Read references/feedback_tagging_guide.md when improving topic/subtopic coverage.
• Read references/live-migration.md when replacing simulated adapters with live sources.