CtrlK
BlogDocsLog inGet started
Tessl Logo

excellence-loop

Applies a 10-point quality rubric to every artifact with iterative refinement until excellence criteria are met. Trigger: "excellence check", "quality rubric", "review artifact", "quality loop", "refine output".

The canonical home for this skill is excellence-loop in JaviMontano/mao-sovereign-architect

SKILL.md
Quality
Evals
Security

Excellence Loop

Applies a rigorous 10-point quality rubric to every deliverable artifact, identifying gaps and iterating until all criteria score above threshold — transforming good work into excellent work.

Guiding Principle

"Excellence is not a destination but a loop. Ship when every dimension scores above threshold, not when you run out of time."

Procedure

Step 1 — Select the Rubric

  1. Load the 10-point excellence rubric appropriate for the artifact type.
  2. For technical documents: completeness, accuracy, evidence, structure, actionability, clarity, diagrams, consistency, audience-fit, next-steps.
  3. For code deliverables: correctness, readability, testability, performance, security, maintainability, documentation, error-handling, naming, standards-compliance.
  4. Customize rubric weights based on artifact priority (critical vs. supporting).
  5. Set the threshold: minimum score per criterion (typically 7/10) and overall (typically 8/10).

Step 2 — Score the Artifact

  1. Evaluate each criterion independently on a 1-10 scale.
  2. Provide specific evidence for each score: what is good, what is lacking.
  3. Identify the lowest-scoring criteria as priority improvement targets.
  4. Calculate the overall weighted score.
  5. Determine the verdict: PASS (all criteria >= threshold), ITERATE (specific criteria below), REWORK (overall below threshold).

Step 3 — Iterate on Gaps

  1. For each criterion below threshold, define the specific improvement needed.
  2. Apply improvements in priority order (lowest scores first).
  3. Re-score only the modified criteria after each improvement pass.
  4. Track iteration count — if more than 3 iterations on the same criterion, escalate for review.
  5. Document what changed in each iteration for audit trail.

Step 4 — Certify and Archive

  1. When all criteria pass, stamp the artifact with the excellence certification.
  2. Record the final scores and iteration history.
  3. Add the ghost menu with next-step actions.
  4. Update the artifact status from {WIP} to ready for review or {Aprobado}.
  5. Feed the scoring data into the lessons-learned tracker for continuous improvement.

Quality Criteria

  • Every artifact is scored against a defined rubric before delivery.
  • No criterion scores below the minimum threshold in the final version.
  • Iteration history is documented showing progressive improvement.
  • Rubric is calibrated quarterly based on feedback and lessons learned.

Anti-Patterns

  • Skipping the excellence loop due to time pressure ("good enough" mentality).
  • Scoring all criteria at 10/10 without honest assessment (rubber stamping).
  • Iterating endlessly on minor improvements (diminishing returns).
  • Using the same rubric for all artifact types without customization.
Repository
JaviMontano/jm-adk
Last updated
First committed

Canonical home

JaviMontano/mao-sovereign-architect
In sync

since Aug 28, 2026

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.