Work Productivity Self Improving Workflow Helper

SKILL.md

---
name: work-productivity-self-improving-workflow-helper
description: Capture lessons from repeated agent or team workflows and turn them into reviewed improvements. Use when the user needs post-run analysis, error pattern capture, memory updates, workflow retrospectives, checklist refinement, or controlled improvement loops.
---

# Work Productivity Self Improving Workflow Helper

Use this skill to convert repeated mistakes, successful interventions, and operational friction into explicit process improvements that are reviewed before being reused.

Read `references/requirement-plan.md` for original demand evidence and review criteria.

## Inputs

Ask for or inspect:

- Recent run logs, failed outputs, review comments, or user corrections.
- The workflow that should improve.
- Existing memory, checklist, prompt, or automation files.
- Constraints on what may be changed automatically.

## Workflow

1. Identify the recurring pattern: error, delay, missing context, bad assumption, or repeated manual step.
2. Separate one-off incidents from reusable lessons.
3. Propose the smallest durable improvement: checklist item, memory note, prompt rule, test, script, or monitoring step.
4. Add review criteria so the improvement can be judged later.
5. Apply edits only to the approved local artifacts; keep live production or publishing steps explicit.
6. Summarize what changed and what future signal should confirm the improvement worked.

## Guardrails

- Do not create self-modifying behavior that bypasses human review.
- Do not preserve sensitive user data in memory unless it is necessary and requested.
- Avoid broad new rules from a single weak example.
- Keep improvement logs concise enough to be useful in future runs.

## Outputs

- Retrospective with root causes and repeated patterns.
- Memory update or checklist patch.
- Proposed workflow change with acceptance criteria.
- Follow-up signal to watch on the next run.

## Validation Checklist

- The improvement is tied to observed evidence.
- The change is small enough to review.
- Sensitive details are omitted or generalized.
- The next run has a concrete way to verify whether the change helped.

## Triggers

Keywords: self-improving agent, lessons learned, retrospective, memory update, repeated failure, workflow improvement, post-run review.

Example requests:

- `Turn these failed runs into durable process improvements.`
- `Use $work-productivity-self-improving-workflow-helper to update the automation memory safely.`
- `Find repeated mistakes in these logs and propose checklist changes.`